TTS models

This document lists all text-to-speech models supported in sherpa-onnx.

Monolingual

The following table lists models by languages.


1-5	Arabic	Catalan	Chinese	Czech	Danish
6-10	Dutch	English	Finnish	French	Georgian
11-15	German	Greek	Hungarian	Icelandic	Italian
16-20	Kazakh	Latvian	Luxembourgish	Malayalam	Nepali
21-25	Norwegian	Persian	Polish	Portuguese	Romanian
26-30	Russian	Serbian	Slovak	Slovenian	Spanish
31-35	Swahili	Swedish	Turkish	Ukrainian	Vietnamese
36	Welsh

Mixed-lingual

The following lists models supporting multiple languages.

Chinese+English

Chinese+English

This section lists text to speech models for Chinese+English.

kokoro-multi-lang-v1_0

Info about this model

This model is kokoro v1.0 and it is from https://huggingface.co/hexgrad/Kokoro-82M

It supports both Chinese and English.

Number of speakers	Sample rate
53	24000

Meaning of speaker prefix

Prefix	Meaning	sid range	Number of speakers
af	American female	0 - 10	11
am	American male	11 - 19	9
bf	British female	20 - 23	4
bm	British male	24 - 27	4
ef	Spanish female	28	1
em	Spanish male	29	1
ff	French female	30	1
hf	Hindi female	31 - 32	2
hm	Hindi male	33 - 34	2
if	Italian female	35	1
im	Italian male	36	1
jf	Japanese female	37 - 40	4
jm	Japanese male	41	1
pf	Brazilian Portuguese female	42	1
pm	Brazilian Portuguese male	43 - 44	2
zf	Chinese female	45 - 48	4
zm	Chinese male	49 - 52	4

speaker ID to speaker name (sid -> name)

The mapping from speaker ID (sid) to speaker name is given below:


0 - 3	0 -> af_alloy	1 -> af_aoede	2 -> af_bella	3 -> af_heart
4 - 7	4 -> af_jessica	5 -> af_kore	6 -> af_nicole	7 -> af_nova
8 - 11	8 -> af_river	9 -> af_sarah	10 -> af_sky	11 -> am_adam
12 - 15	12 -> am_echo	13 -> am_eric	14 -> am_fenrir	15 -> am_liam
16 - 19	16 -> am_michael	17 -> am_onyx	18 -> am_puck	19 -> am_santa
20 - 23	20 -> bf_alice	21 -> bf_emma	22 -> bf_isabella	23 -> bf_lily
24 - 27	24 -> bm_daniel	25 -> bm_fable	26 -> bm_george	27 -> bm_lewis
28 - 31	28 -> ef_dora	29 -> em_alex	30 -> ff_siwis	31 -> hf_alpha
32 - 35	32 -> hf_beta	33 -> hm_omega	34 -> hm_psi	35 -> if_sara
36 - 39	36 -> im_nicola	37 -> jf_alpha	38 -> jf_gongitsune	39 -> jf_nezumi
40 - 43	40 -> jf_tebukuro	41 -> jm_kumo	42 -> pf_dora	43 -> pm_alex
44 - 47	44 -> pm_santa	45 -> zf_xiaobei	46 -> zf_xiaoni	47 -> zf_xiaoxiao
48 - 51	48 -> zf_xiaoyi	49 -> zm_yunjian	50 -> zm_yunxi	51 -> zm_yunxia
52	52 -> zm_yunyang

speaker name to speaker ID (name -> sid)

The mapping from speaker name to speaker ID (sid) is given below:


0 - 3	af_alloy -> 0	af_aoede -> 1	af_bella -> 2	af_heart -> 3
4 - 7	af_jessica -> 4	af_kore -> 5	af_nicole -> 6	af_nova -> 7
8 - 11	af_river -> 8	af_sarah -> 9	af_sky -> 10	am_adam -> 11
12 - 15	am_echo -> 12	am_eric -> 13	am_fenrir -> 14	am_liam -> 15
16 - 19	am_michael -> 16	am_onyx -> 17	am_puck -> 18	am_santa -> 19
20 - 23	bf_alice -> 20	bf_emma -> 21	bf_isabella -> 22	bf_lily -> 23
24 - 27	bm_daniel -> 24	bm_fable -> 25	bm_george -> 26	bm_lewis -> 27
28 - 31	ef_dora -> 28	em_alex -> 29	ff_siwis -> 30	hf_alpha -> 31
32 - 35	hf_beta -> 32	hm_omega -> 33	hm_psi -> 34	if_sara -> 35
36 - 39	im_nicola -> 36	jf_alpha -> 37	jf_gongitsune -> 38	jf_nezumi -> 39
40 - 43	jf_tebukuro -> 40	jm_kumo -> 41	pf_dora -> 42	pm_alex -> 43
44 - 47	pm_santa -> 44	zf_xiaobei -> 45	zf_xiaoni -> 46	zf_xiaoxiao -> 47
48 - 51	zf_xiaoyi -> 48	zm_yunjian -> 49	zm_yunxi -> 50	zm_yunxia -> 51
52 - 52	zm_yunyang -> 52

Samples

For the following text:

This model supports both Chinese and English. 小米的核心价值观是什么？答案
是真诚热爱！有困难，请拨打110 或者18601200909。I am learning 机器学习.
我在研究 machine learning。What do you think 中英文说的如何呢?
今天是 2025年6月18号.

sample audios for different speakers are listed below:

Speaker 0 - af_alloy

Speaker 1 - af_aoede

Speaker 2 - af_bella

Speaker 3 - af_heart

Speaker 4 - af_jessica

Speaker 5 - af_kore

Speaker 6 - af_nicole

Speaker 7 - af_nova

Speaker 8 - af_river

Speaker 9 - af_sarah

Speaker 10 - af_sky

Speaker 11 - am_adam

Speaker 12 - am_echo

Speaker 13 - am_eric

Speaker 14 - am_fenrir

Speaker 15 - am_liam

Speaker 16 - am_michael

Speaker 17 - am_onyx

Speaker 18 - am_puck

Speaker 19 - am_santa

Speaker 20 - bf_alice

Speaker 21 - bf_emma

Speaker 22 - bf_isabella

Speaker 23 - bf_lily

Speaker 24 - bm_daniel

Speaker 25 - bm_fable

Speaker 26 - bm_george

Speaker 27 - bm_lewis

Speaker 28 - ef_dora

Speaker 29 - em_alex

Speaker 30 - ff_siwis

Speaker 31 - hf_alpha

Speaker 32 - hf_beta

Speaker 33 - hm_omega

Speaker 34 - hm_psi

Speaker 35 - if_sara

Speaker 36 - im_nicola

Speaker 37 - jf_alpha

Speaker 38 - jf_gongitsune

Speaker 39 - jf_nezumi

Speaker 40 - jf_tebukuro

Speaker 41 - jm_kumo

Speaker 42 - pf_dora

Speaker 43 - pm_alex

Speaker 44 - pm_santa

Speaker 45 - zf_xiaobei

Speaker 46 - zf_xiaoni

Speaker 47 - zf_xiaoxiao

Speaker 48 - zf_xiaoyi

Speaker 49 - zm_yunjian

Speaker 50 - zm_yunxi

Speaker 51 - zm_yunxia

Speaker 52 - zm_yunyang

kokoro-multi-lang-v1_1

Info about this model

This model is kokoro v1.1-zh and it is from https://huggingface.co/hexgrad/Kokoro-82M-v1.1-zh

It supports both Chinese and English.

Number of speakers	Sample rate
103	24000

Meaning of speaker prefix

Prefix	Meaning	sid range	Number of speakers
af	American female	0 - 1	2
bf	British female	2	1
zf	Chinese female	3 - 57	55
zm	Chinese male	58 - 102	45

speaker ID to speaker name (sid -> name)

The mapping from speaker ID (sid) to speaker name is given below:


0 - 3	0 -> af_maple	1 -> af_sol	2 -> bf_vale	3 -> zf_001
4 - 7	4 -> zf_002	5 -> zf_003	6 -> zf_004	7 -> zf_005
8 - 11	8 -> zf_006	9 -> zf_007	10 -> zf_008	11 -> zf_017
12 - 15	12 -> zf_018	13 -> zf_019	14 -> zf_021	15 -> zf_022
16 - 19	16 -> zf_023	17 -> zf_024	18 -> zf_026	19 -> zf_027
20 - 23	20 -> zf_028	21 -> zf_032	22 -> zf_036	23 -> zf_038
24 - 27	24 -> zf_039	25 -> zf_040	26 -> zf_042	27 -> zf_043
28 - 31	28 -> zf_044	29 -> zf_046	30 -> zf_047	31 -> zf_048
32 - 35	32 -> zf_049	33 -> zf_051	34 -> zf_059	35 -> zf_060
36 - 39	36 -> zf_067	37 -> zf_070	38 -> zf_071	39 -> zf_072
40 - 43	40 -> zf_073	41 -> zf_074	42 -> zf_075	43 -> zf_076
44 - 47	44 -> zf_077	45 -> zf_078	46 -> zf_079	47 -> zf_083
48 - 51	48 -> zf_084	49 -> zf_085	50 -> zf_086	51 -> zf_087
52 - 55	52 -> zf_088	53 -> zf_090	54 -> zf_092	55 -> zf_093
56 - 59	56 -> zf_094	57 -> zf_099	58 -> zm_009	59 -> zm_010
60 - 63	60 -> zm_011	61 -> zm_012	62 -> zm_013	63 -> zm_014
64 - 67	64 -> zm_015	65 -> zm_016	66 -> zm_020	67 -> zm_025
68 - 71	68 -> zm_029	69 -> zm_030	70 -> zm_031	71 -> zm_033
72 - 75	72 -> zm_034	73 -> zm_035	74 -> zm_037	75 -> zm_041
76 - 79	76 -> zm_045	77 -> zm_050	78 -> zm_052	79 -> zm_053
80 - 83	80 -> zm_054	81 -> zm_055	82 -> zm_056	83 -> zm_057
84 - 87	84 -> zm_058	85 -> zm_061	86 -> zm_062	87 -> zm_063
88 - 91	88 -> zm_064	89 -> zm_065	90 -> zm_066	91 -> zm_068
92 - 95	92 -> zm_069	93 -> zm_080	94 -> zm_081	95 -> zm_082
96 - 99	96 -> zm_089	97 -> zm_091	98 -> zm_095	99 -> zm_096
100 - 102	100 -> zm_097	101 -> zm_098	102 -> zm_100

speaker name to speaker ID (name -> sid)

The mapping from speaker name to speaker ID (sid) is given below:


0 - 3	af_maple -> 0	af_sol -> 1	bf_vale -> 2	zf_001 -> 3
4 - 7	zf_002 -> 4	zf_003 -> 5	zf_004 -> 6	zf_005 -> 7
8 - 11	zf_006 -> 8	zf_007 -> 9	zf_008 -> 10	zf_017 -> 11
12 - 15	zf_018 -> 12	zf_019 -> 13	zf_021 -> 14	zf_022 -> 15
16 - 19	zf_023 -> 16	zf_024 -> 17	zf_026 -> 18	zf_027 -> 19
20 - 23	zf_028 -> 20	zf_032 -> 21	zf_036 -> 22	zf_038 -> 23
24 - 27	zf_039 -> 24	zf_040 -> 25	zf_042 -> 26	zf_043 -> 27
28 - 31	zf_044 -> 28	zf_046 -> 29	zf_047 -> 30	zf_048 -> 31
32 - 35	zf_049 -> 32	zf_051 -> 33	zf_059 -> 34	zf_060 -> 35
36 - 39	zf_067 -> 36	zf_070 -> 37	zf_071 -> 38	zf_072 -> 39
40 - 43	zf_073 -> 40	zf_074 -> 41	zf_075 -> 42	zf_076 -> 43
44 - 47	zf_077 -> 44	zf_078 -> 45	zf_079 -> 46	zf_083 -> 47
48 - 51	zf_084 -> 48	zf_085 -> 49	zf_086 -> 50	zf_087 -> 51
52 - 55	zf_088 -> 52	zf_090 -> 53	zf_092 -> 54	zf_093 -> 55
56 - 59	zf_094 -> 56	zf_099 -> 57	zm_009 -> 58	zm_010 -> 59
60 - 63	zm_011 -> 60	zm_012 -> 61	zm_013 -> 62	zm_014 -> 63
64 - 67	zm_015 -> 64	zm_016 -> 65	zm_020 -> 66	zm_025 -> 67
68 - 71	zm_029 -> 68	zm_030 -> 69	zm_031 -> 70	zm_033 -> 71
72 - 75	zm_034 -> 72	zm_035 -> 73	zm_037 -> 74	zm_041 -> 75
76 - 79	zm_045 -> 76	zm_050 -> 77	zm_052 -> 78	zm_053 -> 79
80 - 83	zm_054 -> 80	zm_055 -> 81	zm_056 -> 82	zm_057 -> 83
84 - 87	zm_058 -> 84	zm_061 -> 85	zm_062 -> 86	zm_063 -> 87
88 - 91	zm_064 -> 88	zm_065 -> 89	zm_066 -> 90	zm_068 -> 91
92 - 95	zm_069 -> 92	zm_080 -> 93	zm_081 -> 94	zm_082 -> 95
96 - 99	zm_089 -> 96	zm_091 -> 97	zm_095 -> 98	zm_096 -> 99
100 - 102	zm_097 -> 100	zm_098 -> 101	zm_100 -> 102

Samples

For the following text:

This model supports both Chinese and English. 小米的核心价值观是什么？答案
是真诚热爱！有困难，请拨打110 或者18601200909。I am learning 机器学习.
我在研究 machine learning。What do you think 中英文说的如何呢?
今天是 2025年6月18号.

sample audios for different speakers are listed below:

Speaker 0 - af_maple

Speaker 1 - af_sol

Speaker 2 - bf_vale

Speaker 3 - zf_001

Speaker 4 - zf_002

Speaker 5 - zf_003

Speaker 6 - zf_004

Speaker 7 - zf_005

Speaker 8 - zf_006

Speaker 9 - zf_007

Speaker 10 - zf_008

Speaker 11 - zf_017

Speaker 12 - zf_018

Speaker 13 - zf_019

Speaker 14 - zf_021

Speaker 15 - zf_022

Speaker 16 - zf_023

Speaker 17 - zf_024

Speaker 18 - zf_026

Speaker 19 - zf_027

Speaker 20 - zf_028

Speaker 21 - zf_032

Speaker 22 - zf_036

Speaker 23 - zf_038

Speaker 24 - zf_039

Speaker 25 - zf_040

Speaker 26 - zf_042

Speaker 27 - zf_043

Speaker 28 - zf_044

Speaker 29 - zf_046

Speaker 30 - zf_047

Speaker 31 - zf_048

Speaker 32 - zf_049

Speaker 33 - zf_051

Speaker 34 - zf_059

Speaker 35 - zf_060

Speaker 36 - zf_067

Speaker 37 - zf_070

Speaker 38 - zf_071

Speaker 39 - zf_072

Speaker 40 - zf_073

Speaker 41 - zf_074

Speaker 42 - zf_075

Speaker 43 - zf_076

Speaker 44 - zf_077

Speaker 45 - zf_078

Speaker 46 - zf_079

Speaker 47 - zf_083

Speaker 48 - zf_084

Speaker 49 - zf_085

Speaker 50 - zf_086

Speaker 51 - zf_087

Speaker 52 - zf_088

Speaker 53 - zf_090

Speaker 54 - zf_092

Speaker 55 - zf_093

Speaker 56 - zf_094

Speaker 57 - zf_099

Speaker 58 - zm_009

Speaker 59 - zm_010

Speaker 60 - zm_011

Speaker 61 - zm_012

Speaker 62 - zm_013

Speaker 63 - zm_014

Speaker 64 - zm_015

Speaker 65 - zm_016

Speaker 66 - zm_020

Speaker 67 - zm_025

Speaker 68 - zm_029

Speaker 69 - zm_030

Speaker 70 - zm_031

Speaker 71 - zm_033

Speaker 72 - zm_034

Speaker 73 - zm_035

Speaker 74 - zm_037

Speaker 75 - zm_041

Speaker 76 - zm_045

Speaker 77 - zm_050

Speaker 78 - zm_052

Speaker 79 - zm_053

Speaker 80 - zm_054

Speaker 81 - zm_055

Speaker 82 - zm_056

Speaker 83 - zm_057

Speaker 84 - zm_058

Speaker 85 - zm_061

Speaker 86 - zm_062

Speaker 87 - zm_063

Speaker 88 - zm_064

Speaker 89 - zm_065

Speaker 90 - zm_066

Speaker 91 - zm_068

Speaker 92 - zm_069

Speaker 93 - zm_080

Speaker 94 - zm_081

Speaker 95 - zm_082

Speaker 96 - zm_089

Speaker 97 - zm_091

Speaker 98 - zm_095

Speaker 99 - zm_096

Speaker 100 - zm_097

Speaker 101 - zm_098

Speaker 102 - zm_100

Arabic

This section lists text to speech models for Arabic.

vits-piper-ar_JO-kareem-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ar/ar_JO/kareem/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ar_JO-kareem-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ar_JO-kareem-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ar_JO-kareem-low/ar_JO-kareem-low.onnx";
  config.model.vits.tokens = "vits-piper-ar_JO-kareem-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ar_JO-kareem-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "كيف حالك اليوم؟";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ar_JO-kareem-low.tar.bz2

You can use the following code to play with vits-piper-ar_JO-kareem-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ar_JO-kareem-low/ar_JO-kareem-low.onnx",
            lexicon="",
            data_dir="vits-piper-ar_JO-kareem-low/espeak-ng-data",
            tokens="vits-piper-ar_JO-kareem-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="كيف حالك اليوم؟",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

كيف حالك اليوم؟

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ar_JO-kareem-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ar/ar_JO/kareem/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ar_JO-kareem-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ar_JO-kareem-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ar_JO-kareem-medium/ar_JO-kareem-medium.onnx";
  config.model.vits.tokens = "vits-piper-ar_JO-kareem-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ar_JO-kareem-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "كيف حالك اليوم؟";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ar_JO-kareem-medium.tar.bz2

You can use the following code to play with vits-piper-ar_JO-kareem-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ar_JO-kareem-medium/ar_JO-kareem-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ar_JO-kareem-medium/espeak-ng-data",
            tokens="vits-piper-ar_JO-kareem-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="كيف حالك اليوم؟",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

كيف حالك اليوم؟

sample audios for different speakers are listed below:

Speaker 0

Catalan

This section lists text to speech models for Catalan.

vits-piper-ca_ES-upc_ona-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ca/ca_ES/upc_ona/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ca_ES-upc_ona-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ca_ES-upc_ona-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ca_ES-upc_ona-medium/ca_ES-upc_ona-medium.onnx";
  config.model.vits.tokens = "vits-piper-ca_ES-upc_ona-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ca_ES-upc_ona-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Si vols estar ben servit, fes-te tu mateix el llit";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ca_ES-upc_ona-medium.tar.bz2

You can use the following code to play with vits-piper-ca_ES-upc_ona-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ca_ES-upc_ona-medium/ca_ES-upc_ona-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ca_ES-upc_ona-medium/espeak-ng-data",
            tokens="vits-piper-ca_ES-upc_ona-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Si vols estar ben servit, fes-te tu mateix el llit",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Si vols estar ben servit, fes-te tu mateix el llit

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ca_ES-upc_ona-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ca/ca_ES/upc_ona/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ca_ES-upc_ona-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ca_ES-upc_ona-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ca_ES-upc_ona-x_low/ca_ES-upc_ona-x_low.onnx";
  config.model.vits.tokens = "vits-piper-ca_ES-upc_ona-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ca_ES-upc_ona-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Si vols estar ben servit, fes-te tu mateix el llit";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ca_ES-upc_ona-x_low.tar.bz2

You can use the following code to play with vits-piper-ca_ES-upc_ona-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ca_ES-upc_ona-x_low/ca_ES-upc_ona-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-ca_ES-upc_ona-x_low/espeak-ng-data",
            tokens="vits-piper-ca_ES-upc_ona-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Si vols estar ben servit, fes-te tu mateix el llit",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Si vols estar ben servit, fes-te tu mateix el llit

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ca_ES-upc_pau-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ca/ca_ES/upc_pau/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ca_ES-upc_pau-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ca_ES-upc_pau-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ca_ES-upc_pau-x_low/ca_ES-upc_pau-x_low.onnx";
  config.model.vits.tokens = "vits-piper-ca_ES-upc_pau-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ca_ES-upc_pau-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Si vols estar ben servit, fes-te tu mateix el llit";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ca_ES-upc_pau-x_low.tar.bz2

You can use the following code to play with vits-piper-ca_ES-upc_pau-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ca_ES-upc_pau-x_low/ca_ES-upc_pau-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-ca_ES-upc_pau-x_low/espeak-ng-data",
            tokens="vits-piper-ca_ES-upc_pau-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Si vols estar ben servit, fes-te tu mateix el llit",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Si vols estar ben servit, fes-te tu mateix el llit

sample audios for different speakers are listed below:

Speaker 0

Chinese

This section lists text to speech models for Chinese.

matcha-icefall-zh-baker

matcha-icefall-zh-baker

Info about this model

This model is trained using the code from https://github.com/k2-fsa/icefall/tree/master/egs/baker_zh/TTS/matcha

It supports only Chinese.

Number of speakers	Sample rate
1	22050

Samples

For the following text:

某某银行的副行长和一些行政领导表示，他们去过长江和长白山;
经济不断增长。2024年12月31号，拨打110或者18920240511。123456块钱。
当夜幕降临，星光点点，伴随着微风拂面，我在静谧中感受着时光的流转，
思念如涟漪荡漾，梦境如画卷展开，我与自然融为一体，
沉静在这片宁静的美丽之中，感受着生命的奇迹与温柔.

sample audios for different speakers are listed below:

Speaker 0

Czech

This section lists text to speech models for Czech.

vits-piper-cs_CZ-jirka-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/cs/cs_CZ/jirka/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cs_CZ-jirka-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-cs_CZ-jirka-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-cs_CZ-jirka-low/cs_CZ-jirka-low.onnx";
  config.model.vits.tokens = "vits-piper-cs_CZ-jirka-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-cs_CZ-jirka-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Co můžeš udělat dnes, neodkládej na zítřek. ";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cs_CZ-jirka-low.tar.bz2

You can use the following code to play with vits-piper-cs_CZ-jirka-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-cs_CZ-jirka-low/cs_CZ-jirka-low.onnx",
            lexicon="",
            data_dir="vits-piper-cs_CZ-jirka-low/espeak-ng-data",
            tokens="vits-piper-cs_CZ-jirka-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Co můžeš udělat dnes, neodkládej na zítřek. ",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Co můžeš udělat dnes, neodkládej na zítřek.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-cs_CZ-jirka-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/cs/cs_CZ/jirka/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cs_CZ-jirka-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-cs_CZ-jirka-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-cs_CZ-jirka-medium/cs_CZ-jirka-medium.onnx";
  config.model.vits.tokens = "vits-piper-cs_CZ-jirka-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-cs_CZ-jirka-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Co můžeš udělat dnes, neodkládej na zítřek. ";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cs_CZ-jirka-medium.tar.bz2

You can use the following code to play with vits-piper-cs_CZ-jirka-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-cs_CZ-jirka-medium/cs_CZ-jirka-medium.onnx",
            lexicon="",
            data_dir="vits-piper-cs_CZ-jirka-medium/espeak-ng-data",
            tokens="vits-piper-cs_CZ-jirka-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Co můžeš udělat dnes, neodkládej na zítřek. ",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Co můžeš udělat dnes, neodkládej na zítřek.

sample audios for different speakers are listed below:

Speaker 0

Danish

This section lists text to speech models for Danish.

vits-piper-da_DK-talesyntese-medium

vits-piper-da_DK-talesyntese-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/da/da_DK/talesyntese/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-da_DK-talesyntese-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-da_DK-talesyntese-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-da_DK-talesyntese-medium/da_DK-talesyntese-medium.onnx";
  config.model.vits.tokens = "vits-piper-da_DK-talesyntese-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-da_DK-talesyntese-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Hvis du går langsomt, men aldrig stopper, når du ender frem til dit mål.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-da_DK-talesyntese-medium.tar.bz2

You can use the following code to play with vits-piper-da_DK-talesyntese-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-da_DK-talesyntese-medium/da_DK-talesyntese-medium.onnx",
            lexicon="",
            data_dir="vits-piper-da_DK-talesyntese-medium/espeak-ng-data",
            tokens="vits-piper-da_DK-talesyntese-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Hvis du går langsomt, men aldrig stopper, når du ender frem til dit mål.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Hvis du går langsomt, men aldrig stopper, når du ender frem til dit mål.

sample audios for different speakers are listed below:

Speaker 0

Dutch

This section lists text to speech models for Dutch.

vits-piper-nl_BE-nathalie-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/nl/nl_BE/nathalie/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_BE-nathalie-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-nl_BE-nathalie-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-nl_BE-nathalie-medium/nl_BE-nathalie-medium.onnx";
  config.model.vits.tokens = "vits-piper-nl_BE-nathalie-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-nl_BE-nathalie-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "God schiep het water, maar de Nederlander schiep de dijk";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_BE-nathalie-medium.tar.bz2

You can use the following code to play with vits-piper-nl_BE-nathalie-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-nl_BE-nathalie-medium/nl_BE-nathalie-medium.onnx",
            lexicon="",
            data_dir="vits-piper-nl_BE-nathalie-medium/espeak-ng-data",
            tokens="vits-piper-nl_BE-nathalie-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="God schiep het water, maar de Nederlander schiep de dijk",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

God schiep het water, maar de Nederlander schiep de dijk

sample audios for different speakers are listed below:

Speaker 0

vits-piper-nl_BE-nathalie-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/nl/nl_BE/nathalie/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_BE-nathalie-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-nl_BE-nathalie-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-nl_BE-nathalie-x_low/nl_BE-nathalie-x_low.onnx";
  config.model.vits.tokens = "vits-piper-nl_BE-nathalie-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-nl_BE-nathalie-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "God schiep het water, maar de Nederlander schiep de dijk";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_BE-nathalie-x_low.tar.bz2

You can use the following code to play with vits-piper-nl_BE-nathalie-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-nl_BE-nathalie-x_low/nl_BE-nathalie-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-nl_BE-nathalie-x_low/espeak-ng-data",
            tokens="vits-piper-nl_BE-nathalie-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="God schiep het water, maar de Nederlander schiep de dijk",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

God schiep het water, maar de Nederlander schiep de dijk

sample audios for different speakers are listed below:

Speaker 0

vits-piper-nl_NL-pim-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/nl/nl_NL/pim/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_NL-pim-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-nl_NL-pim-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-nl_NL-pim-medium/nl_NL-pim-medium.onnx";
  config.model.vits.tokens = "vits-piper-nl_NL-pim-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-nl_NL-pim-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "God schiep het water, maar de Nederlander schiep de dijk";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_NL-pim-medium.tar.bz2

You can use the following code to play with vits-piper-nl_NL-pim-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-nl_NL-pim-medium/nl_NL-pim-medium.onnx",
            lexicon="",
            data_dir="vits-piper-nl_NL-pim-medium/espeak-ng-data",
            tokens="vits-piper-nl_NL-pim-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="God schiep het water, maar de Nederlander schiep de dijk",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

God schiep het water, maar de Nederlander schiep de dijk

sample audios for different speakers are listed below:

Speaker 0

vits-piper-nl_NL-ronnie-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/nl/nl_NL/ronnie/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_NL-ronnie-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-nl_NL-ronnie-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-nl_NL-ronnie-medium/nl_NL-ronnie-medium.onnx";
  config.model.vits.tokens = "vits-piper-nl_NL-ronnie-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-nl_NL-ronnie-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "God schiep het water, maar de Nederlander schiep de dijk";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-nl_NL-ronnie-medium.tar.bz2

You can use the following code to play with vits-piper-nl_NL-ronnie-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-nl_NL-ronnie-medium/nl_NL-ronnie-medium.onnx",
            lexicon="",
            data_dir="vits-piper-nl_NL-ronnie-medium/espeak-ng-data",
            tokens="vits-piper-nl_NL-ronnie-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="God schiep het water, maar de Nederlander schiep de dijk",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

God schiep het water, maar de Nederlander schiep de dijk

sample audios for different speakers are listed below:

Speaker 0

English

This section lists text to speech models for English.

kokoro-en-v0_19

Info about this model

This model is kokoro v0.19 and it is from https://huggingface.co/hexgrad/kLegacy

It supports only English.

Number of speakers	Sample rate
11	24000

Meaning of speaker prefix

Prefix	Meaning	sid range	Number of speakers
af	American female	0 - 4	5
am	American male	5 - 6	2
bf	British female	7 - 8	2
bm	British male	9 - 10	2

speaker ID to speaker name (sid -> name)

The mapping from speaker ID (sid) to speaker name is given below:


0 - 3	0 -> af	1 -> af_bella	2 -> af_nicole	3 -> af_sarah
4 - 7	4 -> af_sky	5 -> am_adam	6 -> am_michael	7 -> bf_emma
8 - 10	8 -> bf_isabella	9 -> bm_george	10 -> bm_lewis

speaker name to speaker ID (name -> sid)

The mapping from speaker name to speaker ID (sid) is given below:


0 - 3	af -> 0	af_bella -> 1	af_nicole -> 2	af_sarah -> 3
4 - 7	af_sky -> 4	am_adam -> 5	am_michael -> 6	bf_emma -> 7
8 - 10	bf_isabella -> 8	bm_george -> 9	bm_lewis -> 10

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0 - af

Speaker 1 - af_bella

Speaker 2 - af_nicole

Speaker 3 - af_sarah

Speaker 4 - af_sky

Speaker 5 - am_adam

Speaker 6 - am_michael

Speaker 7 - bf_emma

Speaker 8 - bf_isabella

Speaker 9 - bm_george

Speaker 10 - bm_lewis

vits-piper-en_GB-alan-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/alan/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-alan-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-alan-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-alan-low/en_GB-alan-low.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-alan-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-alan-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-alan-low.tar.bz2

You can use the following code to play with vits-piper-en_GB-alan-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-alan-low/en_GB-alan-low.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-alan-low/espeak-ng-data",
            tokens="vits-piper-en_GB-alan-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-alan-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/alan/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-alan-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-alan-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-alan-medium/en_GB-alan-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-alan-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-alan-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-alan-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-alan-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-alan-medium/en_GB-alan-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-alan-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-alan-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-alba-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/alba/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-alba-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-alba-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-alba-medium/en_GB-alba-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-alba-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-alba-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-alba-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-alba-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-alba-medium/en_GB-alba-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-alba-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-alba-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-aru-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/aru/medium

Number of speakers	Sample rate
12	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-aru-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-aru-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-aru-medium/en_GB-aru-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-aru-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-aru-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-aru-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-aru-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-aru-medium/en_GB-aru-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-aru-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-aru-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

vits-piper-en_GB-cori-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/cori/high

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-cori-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-cori-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-cori-high/en_GB-cori-high.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-cori-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-cori-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-cori-high.tar.bz2

You can use the following code to play with vits-piper-en_GB-cori-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-cori-high/en_GB-cori-high.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-cori-high/espeak-ng-data",
            tokens="vits-piper-en_GB-cori-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-cori-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/cori/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-cori-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-cori-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-cori-medium/en_GB-cori-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-cori-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-cori-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-cori-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-cori-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-cori-medium/en_GB-cori-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-cori-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-cori-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-jenny_dioco-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/jenny_dioco/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-jenny_dioco-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-jenny_dioco-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-jenny_dioco-medium/en_GB-jenny_dioco-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-jenny_dioco-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-jenny_dioco-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-jenny_dioco-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-jenny_dioco-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-jenny_dioco-medium/en_GB-jenny_dioco-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-jenny_dioco-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-jenny_dioco-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-northern_english_male-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/northern_english_male/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-northern_english_male-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-northern_english_male-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-northern_english_male-medium/en_GB-northern_english_male-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-northern_english_male-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-northern_english_male-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-northern_english_male-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-northern_english_male-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-northern_english_male-medium/en_GB-northern_english_male-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-northern_english_male-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-northern_english_male-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-semaine-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/semaine/medium

Number of speakers	Sample rate
4	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-semaine-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-semaine-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-semaine-medium/en_GB-semaine-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-semaine-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-semaine-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-semaine-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-semaine-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-semaine-medium/en_GB-semaine-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-semaine-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-semaine-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

vits-piper-en_GB-southern_english_female-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/southern_english_female/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-southern_english_female-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-southern_english_female-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-southern_english_female-low/en_GB-southern_english_female-low.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-southern_english_female-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-southern_english_female-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-southern_english_female-low.tar.bz2

You can use the following code to play with vits-piper-en_GB-southern_english_female-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-southern_english_female-low/en_GB-southern_english_female-low.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-southern_english_female-low/espeak-ng-data",
            tokens="vits-piper-en_GB-southern_english_female-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_GB-southern_english_female-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/csukuangfj/vits-piper-en_GB-southern_english_female-medium

Number of speakers	Sample rate
6	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-southern_english_female-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-southern_english_female-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-southern_english_female-medium/en_GB-southern_english_female-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-southern_english_female-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-southern_english_female-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-southern_english_female-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-southern_english_female-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-southern_english_female-medium/en_GB-southern_english_female-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-southern_english_female-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-southern_english_female-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

vits-piper-en_GB-southern_english_male-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/csukuangfj/vits-piper-en_GB-southern_english_male-medium

Number of speakers	Sample rate
8	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-southern_english_male-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-southern_english_male-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-southern_english_male-medium/en_GB-southern_english_male-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-southern_english_male-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-southern_english_male-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-southern_english_male-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-southern_english_male-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-southern_english_male-medium/en_GB-southern_english_male-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-southern_english_male-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-southern_english_male-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

vits-piper-en_GB-vctk-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_GB/vctk/medium

Number of speakers	Sample rate
109	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-vctk-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_GB-vctk-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_GB-vctk-medium/en_GB-vctk-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_GB-vctk-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_GB-vctk-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-vctk-medium.tar.bz2

You can use the following code to play with vits-piper-en_GB-vctk-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_GB-vctk-medium/en_GB-vctk-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_GB-vctk-medium/espeak-ng-data",
            tokens="vits-piper-en_GB-vctk-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

Speaker 18

Speaker 19

Speaker 20

Speaker 21

Speaker 22

Speaker 23

Speaker 24

Speaker 25

Speaker 26

Speaker 27

Speaker 28

Speaker 29

Speaker 30

Speaker 31

Speaker 32

Speaker 33

Speaker 34

Speaker 35

Speaker 36

Speaker 37

Speaker 38

Speaker 39

Speaker 40

Speaker 41

Speaker 42

Speaker 43

Speaker 44

Speaker 45

Speaker 46

Speaker 47

Speaker 48

Speaker 49

Speaker 50

Speaker 51

Speaker 52

Speaker 53

Speaker 54

Speaker 55

Speaker 56

Speaker 57

Speaker 58

Speaker 59

Speaker 60

Speaker 61

Speaker 62

Speaker 63

Speaker 64

Speaker 65

Speaker 66

Speaker 67

Speaker 68

Speaker 69

Speaker 70

Speaker 71

Speaker 72

Speaker 73

Speaker 74

Speaker 75

Speaker 76

Speaker 77

Speaker 78

Speaker 79

Speaker 80

Speaker 81

Speaker 82

Speaker 83

Speaker 84

Speaker 85

Speaker 86

Speaker 87

Speaker 88

Speaker 89

Speaker 90

Speaker 91

Speaker 92

Speaker 93

Speaker 94

Speaker 95

Speaker 96

Speaker 97

Speaker 98

Speaker 99

Speaker 100

Speaker 101

Speaker 102

Speaker 103

Speaker 104

Speaker 105

Speaker 106

Speaker 107

Speaker 108

vits-piper-en_US-amy-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/amy/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-amy-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-amy-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-amy-low/en_US-amy-low.onnx";
  config.model.vits.tokens = "vits-piper-en_US-amy-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-amy-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-amy-low.tar.bz2

You can use the following code to play with vits-piper-en_US-amy-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-amy-low/en_US-amy-low.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-amy-low/espeak-ng-data",
            tokens="vits-piper-en_US-amy-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-amy-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/amy/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-amy-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-amy-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-amy-medium/en_US-amy-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-amy-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-amy-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-amy-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-amy-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-amy-medium/en_US-amy-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-amy-medium/espeak-ng-data",
            tokens="vits-piper-en_US-amy-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-arctic-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/arctic/medium

Number of speakers	Sample rate
18	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-arctic-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-arctic-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-arctic-medium/en_US-arctic-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-arctic-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-arctic-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-arctic-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-arctic-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-arctic-medium/en_US-arctic-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-arctic-medium/espeak-ng-data",
            tokens="vits-piper-en_US-arctic-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

vits-piper-en_US-bryce-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/bryce/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-bryce-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-bryce-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-bryce-medium/en_US-bryce-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-bryce-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-bryce-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-bryce-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-bryce-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-bryce-medium/en_US-bryce-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-bryce-medium/espeak-ng-data",
            tokens="vits-piper-en_US-bryce-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-danny-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/danny/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-danny-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-danny-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-danny-low/en_US-danny-low.onnx";
  config.model.vits.tokens = "vits-piper-en_US-danny-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-danny-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-danny-low.tar.bz2

You can use the following code to play with vits-piper-en_US-danny-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-danny-low/en_US-danny-low.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-danny-low/espeak-ng-data",
            tokens="vits-piper-en_US-danny-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-glados-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://github.com/rhasspy/piper/issues/187#issuecomment-1805709037

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-glados-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-glados-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-glados-high/en_US-glados-high.onnx";
  config.model.vits.tokens = "vits-piper-en_US-glados-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-glados-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-glados-high.tar.bz2

You can use the following code to play with vits-piper-en_US-glados-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-glados-high/en_US-glados-high.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-glados-high/espeak-ng-data",
            tokens="vits-piper-en_US-glados-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-hfc_female-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/hfc_female/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-hfc_female-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-hfc_female-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-hfc_female-medium/en_US-hfc_female-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-hfc_female-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-hfc_female-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-hfc_female-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-hfc_female-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-hfc_female-medium/en_US-hfc_female-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-hfc_female-medium/espeak-ng-data",
            tokens="vits-piper-en_US-hfc_female-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-hfc_male-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/hfc_male/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-hfc_male-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-hfc_male-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-hfc_male-medium/en_US-hfc_male-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-hfc_male-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-hfc_male-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-hfc_male-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-hfc_male-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-hfc_male-medium/en_US-hfc_male-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-hfc_male-medium/espeak-ng-data",
            tokens="vits-piper-en_US-hfc_male-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-joe-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/joe/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-joe-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-joe-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-joe-medium/en_US-joe-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-joe-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-joe-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-joe-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-joe-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-joe-medium/en_US-joe-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-joe-medium/espeak-ng-data",
            tokens="vits-piper-en_US-joe-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-john-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/john/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-john-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-john-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-john-medium/en_US-john-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-john-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-john-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-john-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-john-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-john-medium/en_US-john-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-john-medium/espeak-ng-data",
            tokens="vits-piper-en_US-john-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-kathleen-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/kathleen/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-kathleen-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-kathleen-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-kathleen-low/en_US-kathleen-low.onnx";
  config.model.vits.tokens = "vits-piper-en_US-kathleen-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-kathleen-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-kathleen-low.tar.bz2

You can use the following code to play with vits-piper-en_US-kathleen-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-kathleen-low/en_US-kathleen-low.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-kathleen-low/espeak-ng-data",
            tokens="vits-piper-en_US-kathleen-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-kristin-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/kristin/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-kristin-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-kristin-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-kristin-medium/en_US-kristin-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-kristin-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-kristin-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-kristin-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-kristin-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-kristin-medium/en_US-kristin-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-kristin-medium/espeak-ng-data",
            tokens="vits-piper-en_US-kristin-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-kusal-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/kusal/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-kusal-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-kusal-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-kusal-medium/en_US-kusal-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-kusal-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-kusal-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-kusal-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-kusal-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-kusal-medium/en_US-kusal-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-kusal-medium/espeak-ng-data",
            tokens="vits-piper-en_US-kusal-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-l2arctic-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/l2arctic/medium

Number of speakers	Sample rate
24	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-l2arctic-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-l2arctic-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-l2arctic-medium/en_US-l2arctic-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-l2arctic-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-l2arctic-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-l2arctic-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-l2arctic-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-l2arctic-medium/en_US-l2arctic-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-l2arctic-medium/espeak-ng-data",
            tokens="vits-piper-en_US-l2arctic-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

Speaker 18

Speaker 19

Speaker 20

Speaker 21

Speaker 22

Speaker 23

vits-piper-en_US-lessac-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/lessac/high

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-lessac-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-lessac-high/en_US-lessac-high.onnx";
  config.model.vits.tokens = "vits-piper-en_US-lessac-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-lessac-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-high.tar.bz2

You can use the following code to play with vits-piper-en_US-lessac-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-lessac-high/en_US-lessac-high.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-lessac-high/espeak-ng-data",
            tokens="vits-piper-en_US-lessac-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-lessac-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/lessac/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-lessac-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-lessac-low/en_US-lessac-low.onnx";
  config.model.vits.tokens = "vits-piper-en_US-lessac-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-lessac-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-low.tar.bz2

You can use the following code to play with vits-piper-en_US-lessac-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-lessac-low/en_US-lessac-low.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-lessac-low/espeak-ng-data",
            tokens="vits-piper-en_US-lessac-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-lessac-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/lessac/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-lessac-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-lessac-medium/en_US-lessac-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-lessac-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-lessac-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-lessac-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-lessac-medium/en_US-lessac-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-lessac-medium/espeak-ng-data",
            tokens="vits-piper-en_US-lessac-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-libritts-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/libritts/high

Number of speakers	Sample rate
904	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-libritts-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-libritts-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-libritts-high/en_US-libritts-high.onnx";
  config.model.vits.tokens = "vits-piper-en_US-libritts-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-libritts-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-libritts-high.tar.bz2

You can use the following code to play with vits-piper-en_US-libritts-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-libritts-high/en_US-libritts-high.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-libritts-high/espeak-ng-data",
            tokens="vits-piper-en_US-libritts-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

Speaker 18

Speaker 19

Speaker 20

Speaker 21

Speaker 22

Speaker 23

Speaker 24

Speaker 25

Speaker 26

Speaker 27

Speaker 28

Speaker 29

Speaker 30

Speaker 31

Speaker 32

Speaker 33

Speaker 34

Speaker 35

Speaker 36

Speaker 37

Speaker 38

Speaker 39

Speaker 40

Speaker 41

Speaker 42

Speaker 43

Speaker 44

Speaker 45

Speaker 46

Speaker 47

Speaker 48

Speaker 49

Speaker 50

Speaker 51

Speaker 52

Speaker 53

Speaker 54

Speaker 55

Speaker 56

Speaker 57

Speaker 58

Speaker 59

Speaker 60

Speaker 61

Speaker 62

Speaker 63

Speaker 64

Speaker 65

Speaker 66

Speaker 67

Speaker 68

Speaker 69

Speaker 70

Speaker 71

Speaker 72

Speaker 73

Speaker 74

Speaker 75

Speaker 76

Speaker 77

Speaker 78

Speaker 79

Speaker 80

Speaker 81

Speaker 82

Speaker 83

Speaker 84

Speaker 85

Speaker 86

Speaker 87

Speaker 88

Speaker 89

Speaker 90

Speaker 91

Speaker 92

Speaker 93

Speaker 94

Speaker 95

Speaker 96

Speaker 97

Speaker 98

Speaker 99

Speaker 100

Speaker 101

Speaker 102

Speaker 103

Speaker 104

Speaker 105

Speaker 106

Speaker 107

Speaker 108

Speaker 109

Speaker 110

Speaker 111

Speaker 112

Speaker 113

Speaker 114

Speaker 115

Speaker 116

Speaker 117

Speaker 118

Speaker 119

Speaker 120

Speaker 121

Speaker 122

Speaker 123

Speaker 124

Speaker 125

Speaker 126

Speaker 127

Speaker 128

Speaker 129

Speaker 130

Speaker 131

Speaker 132

Speaker 133

Speaker 134

Speaker 135

Speaker 136

Speaker 137

Speaker 138

Speaker 139

Speaker 140

Speaker 141

Speaker 142

Speaker 143

Speaker 144

Speaker 145

Speaker 146

Speaker 147

Speaker 148

Speaker 149

Speaker 150

Speaker 151

Speaker 152

Speaker 153

Speaker 154

Speaker 155

Speaker 156

Speaker 157

Speaker 158

Speaker 159

Speaker 160

Speaker 161

Speaker 162

Speaker 163

Speaker 164

Speaker 165

Speaker 166

Speaker 167

Speaker 168

Speaker 169

Speaker 170

Speaker 171

Speaker 172

Speaker 173

Speaker 174

Speaker 175

Speaker 176

Speaker 177

Speaker 178

Speaker 179

Speaker 180

Speaker 181

Speaker 182

Speaker 183

Speaker 184

Speaker 185

Speaker 186

Speaker 187

Speaker 188

Speaker 189

Speaker 190

Speaker 191

Speaker 192

Speaker 193

Speaker 194

Speaker 195

Speaker 196

Speaker 197

Speaker 198

Speaker 199

Speaker 200

Speaker 201

Speaker 202

Speaker 203

Speaker 204

Speaker 205

Speaker 206

Speaker 207

Speaker 208

Speaker 209

Speaker 210

Speaker 211

Speaker 212

Speaker 213

Speaker 214

Speaker 215

Speaker 216

Speaker 217

Speaker 218

Speaker 219

Speaker 220

Speaker 221

Speaker 222

Speaker 223

Speaker 224

Speaker 225

Speaker 226

Speaker 227

Speaker 228

Speaker 229

Speaker 230

Speaker 231

Speaker 232

Speaker 233

Speaker 234

Speaker 235

Speaker 236

Speaker 237

Speaker 238

Speaker 239

Speaker 240

Speaker 241

Speaker 242

Speaker 243

Speaker 244

Speaker 245

Speaker 246

Speaker 247

Speaker 248

Speaker 249

Speaker 250

Speaker 251

Speaker 252

Speaker 253

Speaker 254

Speaker 255

Speaker 256

Speaker 257

Speaker 258

Speaker 259

Speaker 260

Speaker 261

Speaker 262

Speaker 263

Speaker 264

Speaker 265

Speaker 266

Speaker 267

Speaker 268

Speaker 269

Speaker 270

Speaker 271

Speaker 272

Speaker 273

Speaker 274

Speaker 275

Speaker 276

Speaker 277

Speaker 278

Speaker 279

Speaker 280

Speaker 281

Speaker 282

Speaker 283

Speaker 284

Speaker 285

Speaker 286

Speaker 287

Speaker 288

Speaker 289

Speaker 290

Speaker 291

Speaker 292

Speaker 293

Speaker 294

Speaker 295

Speaker 296

Speaker 297

Speaker 298

Speaker 299

Speaker 300

Speaker 301

Speaker 302

Speaker 303

Speaker 304

Speaker 305

Speaker 306

Speaker 307

Speaker 308

Speaker 309

Speaker 310

Speaker 311

Speaker 312

Speaker 313

Speaker 314

Speaker 315

Speaker 316

Speaker 317

Speaker 318

Speaker 319

Speaker 320

Speaker 321

Speaker 322

Speaker 323

Speaker 324

Speaker 325

Speaker 326

Speaker 327

Speaker 328

Speaker 329

Speaker 330

Speaker 331

Speaker 332

Speaker 333

Speaker 334

Speaker 335

Speaker 336

Speaker 337

Speaker 338

Speaker 339

Speaker 340

Speaker 341

Speaker 342

Speaker 343

Speaker 344

Speaker 345

Speaker 346

Speaker 347

Speaker 348

Speaker 349

Speaker 350

Speaker 351

Speaker 352

Speaker 353

Speaker 354

Speaker 355

Speaker 356

Speaker 357

Speaker 358

Speaker 359

Speaker 360

Speaker 361

Speaker 362

Speaker 363

Speaker 364

Speaker 365

Speaker 366

Speaker 367

Speaker 368

Speaker 369

Speaker 370

Speaker 371

Speaker 372

Speaker 373

Speaker 374

Speaker 375

Speaker 376

Speaker 377

Speaker 378

Speaker 379

Speaker 380

Speaker 381

Speaker 382

Speaker 383

Speaker 384

Speaker 385

Speaker 386

Speaker 387

Speaker 388

Speaker 389

Speaker 390

Speaker 391

Speaker 392

Speaker 393

Speaker 394

Speaker 395

Speaker 396

Speaker 397

Speaker 398

Speaker 399

Speaker 400

Speaker 401

Speaker 402

Speaker 403

Speaker 404

Speaker 405

Speaker 406

Speaker 407

Speaker 408

Speaker 409

Speaker 410

Speaker 411

Speaker 412

Speaker 413

Speaker 414

Speaker 415

Speaker 416

Speaker 417

Speaker 418

Speaker 419

Speaker 420

Speaker 421

Speaker 422

Speaker 423

Speaker 424

Speaker 425

Speaker 426

Speaker 427

Speaker 428

Speaker 429

Speaker 430

Speaker 431

Speaker 432

Speaker 433

Speaker 434

Speaker 435

Speaker 436

Speaker 437

Speaker 438

Speaker 439

Speaker 440

Speaker 441

Speaker 442

Speaker 443

Speaker 444

Speaker 445

Speaker 446

Speaker 447

Speaker 448

Speaker 449

Speaker 450

Speaker 451

Speaker 452

Speaker 453

Speaker 454

Speaker 455

Speaker 456

Speaker 457

Speaker 458

Speaker 459

Speaker 460

Speaker 461

Speaker 462

Speaker 463

Speaker 464

Speaker 465

Speaker 466

Speaker 467

Speaker 468

Speaker 469

Speaker 470

Speaker 471

Speaker 472

Speaker 473

Speaker 474

Speaker 475

Speaker 476

Speaker 477

Speaker 478

Speaker 479

Speaker 480

Speaker 481

Speaker 482

Speaker 483

Speaker 484

Speaker 485

Speaker 486

Speaker 487

Speaker 488

Speaker 489

Speaker 490

Speaker 491

Speaker 492

Speaker 493

Speaker 494

Speaker 495

Speaker 496

Speaker 497

Speaker 498

Speaker 499

Speaker 500

Speaker 501

Speaker 502

Speaker 503

Speaker 504

Speaker 505

Speaker 506

Speaker 507

Speaker 508

Speaker 509

Speaker 510

Speaker 511

Speaker 512

Speaker 513

Speaker 514

Speaker 515

Speaker 516

Speaker 517

Speaker 518

Speaker 519

Speaker 520

Speaker 521

Speaker 522

Speaker 523

Speaker 524

Speaker 525

Speaker 526

Speaker 527

Speaker 528

Speaker 529

Speaker 530

Speaker 531

Speaker 532

Speaker 533

Speaker 534

Speaker 535

Speaker 536

Speaker 537

Speaker 538

Speaker 539

Speaker 540

Speaker 541

Speaker 542

Speaker 543

Speaker 544

Speaker 545

Speaker 546

Speaker 547

Speaker 548

Speaker 549

Speaker 550

Speaker 551

Speaker 552

Speaker 553

Speaker 554

Speaker 555

Speaker 556

Speaker 557

Speaker 558

Speaker 559

Speaker 560

Speaker 561

Speaker 562

Speaker 563

Speaker 564

Speaker 565

Speaker 566

Speaker 567

Speaker 568

Speaker 569

Speaker 570

Speaker 571

Speaker 572

Speaker 573

Speaker 574

Speaker 575

Speaker 576

Speaker 577

Speaker 578

Speaker 579

Speaker 580

Speaker 581

Speaker 582

Speaker 583

Speaker 584

Speaker 585

Speaker 586

Speaker 587

Speaker 588

Speaker 589

Speaker 590

Speaker 591

Speaker 592

Speaker 593

Speaker 594

Speaker 595

Speaker 596

Speaker 597

Speaker 598

Speaker 599

Speaker 600

Speaker 601

Speaker 602

Speaker 603

Speaker 604

Speaker 605

Speaker 606

Speaker 607

Speaker 608

Speaker 609

Speaker 610

Speaker 611

Speaker 612

Speaker 613

Speaker 614

Speaker 615

Speaker 616

Speaker 617

Speaker 618

Speaker 619

Speaker 620

Speaker 621

Speaker 622

Speaker 623

Speaker 624

Speaker 625

Speaker 626

Speaker 627

Speaker 628

Speaker 629

Speaker 630

Speaker 631

Speaker 632

Speaker 633

Speaker 634

Speaker 635

Speaker 636

Speaker 637

Speaker 638

Speaker 639

Speaker 640

Speaker 641

Speaker 642

Speaker 643

Speaker 644

Speaker 645

Speaker 646

Speaker 647

Speaker 648

Speaker 649

Speaker 650

Speaker 651

Speaker 652

Speaker 653

Speaker 654

Speaker 655

Speaker 656

Speaker 657

Speaker 658

Speaker 659

Speaker 660

Speaker 661

Speaker 662

Speaker 663

Speaker 664

Speaker 665

Speaker 666

Speaker 667

Speaker 668

Speaker 669

Speaker 670

Speaker 671

Speaker 672

Speaker 673

Speaker 674

Speaker 675

Speaker 676

Speaker 677

Speaker 678

Speaker 679

Speaker 680

Speaker 681

Speaker 682

Speaker 683

Speaker 684

Speaker 685

Speaker 686

Speaker 687

Speaker 688

Speaker 689

Speaker 690

Speaker 691

Speaker 692

Speaker 693

Speaker 694

Speaker 695

Speaker 696

Speaker 697

Speaker 698

Speaker 699

Speaker 700

Speaker 701

Speaker 702

Speaker 703

Speaker 704

Speaker 705

Speaker 706

Speaker 707

Speaker 708

Speaker 709

Speaker 710

Speaker 711

Speaker 712

Speaker 713

Speaker 714

Speaker 715

Speaker 716

Speaker 717

Speaker 718

Speaker 719

Speaker 720

Speaker 721

Speaker 722

Speaker 723

Speaker 724

Speaker 725

Speaker 726

Speaker 727

Speaker 728

Speaker 729

Speaker 730

Speaker 731

Speaker 732

Speaker 733

Speaker 734

Speaker 735

Speaker 736

Speaker 737

Speaker 738

Speaker 739

Speaker 740

Speaker 741

Speaker 742

Speaker 743

Speaker 744

Speaker 745

Speaker 746

Speaker 747

Speaker 748

Speaker 749

Speaker 750

Speaker 751

Speaker 752

Speaker 753

Speaker 754

Speaker 755

Speaker 756

Speaker 757

Speaker 758

Speaker 759

Speaker 760

Speaker 761

Speaker 762

Speaker 763

Speaker 764

Speaker 765

Speaker 766

Speaker 767

Speaker 768

Speaker 769

Speaker 770

Speaker 771

Speaker 772

Speaker 773

Speaker 774

Speaker 775

Speaker 776

Speaker 777

Speaker 778

Speaker 779

Speaker 780

Speaker 781

Speaker 782

Speaker 783

Speaker 784

Speaker 785

Speaker 786

Speaker 787

Speaker 788

Speaker 789

Speaker 790

Speaker 791

Speaker 792

Speaker 793

Speaker 794

Speaker 795

Speaker 796

Speaker 797

Speaker 798

Speaker 799

Speaker 800

Speaker 801

Speaker 802

Speaker 803

Speaker 804

Speaker 805

Speaker 806

Speaker 807

Speaker 808

Speaker 809

Speaker 810

Speaker 811

Speaker 812

Speaker 813

Speaker 814

Speaker 815

Speaker 816

Speaker 817

Speaker 818

Speaker 819

Speaker 820

Speaker 821

Speaker 822

Speaker 823

Speaker 824

Speaker 825

Speaker 826

Speaker 827

Speaker 828

Speaker 829

Speaker 830

Speaker 831

Speaker 832

Speaker 833

Speaker 834

Speaker 835

Speaker 836

Speaker 837

Speaker 838

Speaker 839

Speaker 840

Speaker 841

Speaker 842

Speaker 843

Speaker 844

Speaker 845

Speaker 846

Speaker 847

Speaker 848

Speaker 849

Speaker 850

Speaker 851

Speaker 852

Speaker 853

Speaker 854

Speaker 855

Speaker 856

Speaker 857

Speaker 858

Speaker 859

Speaker 860

Speaker 861

Speaker 862

Speaker 863

Speaker 864

Speaker 865

Speaker 866

Speaker 867

Speaker 868

Speaker 869

Speaker 870

Speaker 871

Speaker 872

Speaker 873

Speaker 874

Speaker 875

Speaker 876

Speaker 877

Speaker 878

Speaker 879

Speaker 880

Speaker 881

Speaker 882

Speaker 883

Speaker 884

Speaker 885

Speaker 886

Speaker 887

Speaker 888

Speaker 889

Speaker 890

Speaker 891

Speaker 892

Speaker 893

Speaker 894

Speaker 895

Speaker 896

Speaker 897

Speaker 898

Speaker 899

Speaker 900

Speaker 901

Speaker 902

Speaker 903

vits-piper-en_US-libritts_r-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/libritts_r/medium

Number of speakers	Sample rate
904	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-libritts_r-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-libritts_r-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-libritts_r-medium/en_US-libritts_r-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-libritts_r-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-libritts_r-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-libritts_r-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-libritts_r-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-libritts_r-medium/en_US-libritts_r-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-libritts_r-medium/espeak-ng-data",
            tokens="vits-piper-en_US-libritts_r-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

Speaker 18

Speaker 19

Speaker 20

Speaker 21

Speaker 22

Speaker 23

Speaker 24

Speaker 25

Speaker 26

Speaker 27

Speaker 28

Speaker 29

Speaker 30

Speaker 31

Speaker 32

Speaker 33

Speaker 34

Speaker 35

Speaker 36

Speaker 37

Speaker 38

Speaker 39

Speaker 40

Speaker 41

Speaker 42

Speaker 43

Speaker 44

Speaker 45

Speaker 46

Speaker 47

Speaker 48

Speaker 49

Speaker 50

Speaker 51

Speaker 52

Speaker 53

Speaker 54

Speaker 55

Speaker 56

Speaker 57

Speaker 58

Speaker 59

Speaker 60

Speaker 61

Speaker 62

Speaker 63

Speaker 64

Speaker 65

Speaker 66

Speaker 67

Speaker 68

Speaker 69

Speaker 70

Speaker 71

Speaker 72

Speaker 73

Speaker 74

Speaker 75

Speaker 76

Speaker 77

Speaker 78

Speaker 79

Speaker 80

Speaker 81

Speaker 82

Speaker 83

Speaker 84

Speaker 85

Speaker 86

Speaker 87

Speaker 88

Speaker 89

Speaker 90

Speaker 91

Speaker 92

Speaker 93

Speaker 94

Speaker 95

Speaker 96

Speaker 97

Speaker 98

Speaker 99

Speaker 100

Speaker 101

Speaker 102

Speaker 103

Speaker 104

Speaker 105

Speaker 106

Speaker 107

Speaker 108

Speaker 109

Speaker 110

Speaker 111

Speaker 112

Speaker 113

Speaker 114

Speaker 115

Speaker 116

Speaker 117

Speaker 118

Speaker 119

Speaker 120

Speaker 121

Speaker 122

Speaker 123

Speaker 124

Speaker 125

Speaker 126

Speaker 127

Speaker 128

Speaker 129

Speaker 130

Speaker 131

Speaker 132

Speaker 133

Speaker 134

Speaker 135

Speaker 136

Speaker 137

Speaker 138

Speaker 139

Speaker 140

Speaker 141

Speaker 142

Speaker 143

Speaker 144

Speaker 145

Speaker 146

Speaker 147

Speaker 148

Speaker 149

Speaker 150

Speaker 151

Speaker 152

Speaker 153

Speaker 154

Speaker 155

Speaker 156

Speaker 157

Speaker 158

Speaker 159

Speaker 160

Speaker 161

Speaker 162

Speaker 163

Speaker 164

Speaker 165

Speaker 166

Speaker 167

Speaker 168

Speaker 169

Speaker 170

Speaker 171

Speaker 172

Speaker 173

Speaker 174

Speaker 175

Speaker 176

Speaker 177

Speaker 178

Speaker 179

Speaker 180

Speaker 181

Speaker 182

Speaker 183

Speaker 184

Speaker 185

Speaker 186

Speaker 187

Speaker 188

Speaker 189

Speaker 190

Speaker 191

Speaker 192

Speaker 193

Speaker 194

Speaker 195

Speaker 196

Speaker 197

Speaker 198

Speaker 199

Speaker 200

Speaker 201

Speaker 202

Speaker 203

Speaker 204

Speaker 205

Speaker 206

Speaker 207

Speaker 208

Speaker 209

Speaker 210

Speaker 211

Speaker 212

Speaker 213

Speaker 214

Speaker 215

Speaker 216

Speaker 217

Speaker 218

Speaker 219

Speaker 220

Speaker 221

Speaker 222

Speaker 223

Speaker 224

Speaker 225

Speaker 226

Speaker 227

Speaker 228

Speaker 229

Speaker 230

Speaker 231

Speaker 232

Speaker 233

Speaker 234

Speaker 235

Speaker 236

Speaker 237

Speaker 238

Speaker 239

Speaker 240

Speaker 241

Speaker 242

Speaker 243

Speaker 244

Speaker 245

Speaker 246

Speaker 247

Speaker 248

Speaker 249

Speaker 250

Speaker 251

Speaker 252

Speaker 253

Speaker 254

Speaker 255

Speaker 256

Speaker 257

Speaker 258

Speaker 259

Speaker 260

Speaker 261

Speaker 262

Speaker 263

Speaker 264

Speaker 265

Speaker 266

Speaker 267

Speaker 268

Speaker 269

Speaker 270

Speaker 271

Speaker 272

Speaker 273

Speaker 274

Speaker 275

Speaker 276

Speaker 277

Speaker 278

Speaker 279

Speaker 280

Speaker 281

Speaker 282

Speaker 283

Speaker 284

Speaker 285

Speaker 286

Speaker 287

Speaker 288

Speaker 289

Speaker 290

Speaker 291

Speaker 292

Speaker 293

Speaker 294

Speaker 295

Speaker 296

Speaker 297

Speaker 298

Speaker 299

Speaker 300

Speaker 301

Speaker 302

Speaker 303

Speaker 304

Speaker 305

Speaker 306

Speaker 307

Speaker 308

Speaker 309

Speaker 310

Speaker 311

Speaker 312

Speaker 313

Speaker 314

Speaker 315

Speaker 316

Speaker 317

Speaker 318

Speaker 319

Speaker 320

Speaker 321

Speaker 322

Speaker 323

Speaker 324

Speaker 325

Speaker 326

Speaker 327

Speaker 328

Speaker 329

Speaker 330

Speaker 331

Speaker 332

Speaker 333

Speaker 334

Speaker 335

Speaker 336

Speaker 337

Speaker 338

Speaker 339

Speaker 340

Speaker 341

Speaker 342

Speaker 343

Speaker 344

Speaker 345

Speaker 346

Speaker 347

Speaker 348

Speaker 349

Speaker 350

Speaker 351

Speaker 352

Speaker 353

Speaker 354

Speaker 355

Speaker 356

Speaker 357

Speaker 358

Speaker 359

Speaker 360

Speaker 361

Speaker 362

Speaker 363

Speaker 364

Speaker 365

Speaker 366

Speaker 367

Speaker 368

Speaker 369

Speaker 370

Speaker 371

Speaker 372

Speaker 373

Speaker 374

Speaker 375

Speaker 376

Speaker 377

Speaker 378

Speaker 379

Speaker 380

Speaker 381

Speaker 382

Speaker 383

Speaker 384

Speaker 385

Speaker 386

Speaker 387

Speaker 388

Speaker 389

Speaker 390

Speaker 391

Speaker 392

Speaker 393

Speaker 394

Speaker 395

Speaker 396

Speaker 397

Speaker 398

Speaker 399

Speaker 400

Speaker 401

Speaker 402

Speaker 403

Speaker 404

Speaker 405

Speaker 406

Speaker 407

Speaker 408

Speaker 409

Speaker 410

Speaker 411

Speaker 412

Speaker 413

Speaker 414

Speaker 415

Speaker 416

Speaker 417

Speaker 418

Speaker 419

Speaker 420

Speaker 421

Speaker 422

Speaker 423

Speaker 424

Speaker 425

Speaker 426

Speaker 427

Speaker 428

Speaker 429

Speaker 430

Speaker 431

Speaker 432

Speaker 433

Speaker 434

Speaker 435

Speaker 436

Speaker 437

Speaker 438

Speaker 439

Speaker 440

Speaker 441

Speaker 442

Speaker 443

Speaker 444

Speaker 445

Speaker 446

Speaker 447

Speaker 448

Speaker 449

Speaker 450

Speaker 451

Speaker 452

Speaker 453

Speaker 454

Speaker 455

Speaker 456

Speaker 457

Speaker 458

Speaker 459

Speaker 460

Speaker 461

Speaker 462

Speaker 463

Speaker 464

Speaker 465

Speaker 466

Speaker 467

Speaker 468

Speaker 469

Speaker 470

Speaker 471

Speaker 472

Speaker 473

Speaker 474

Speaker 475

Speaker 476

Speaker 477

Speaker 478

Speaker 479

Speaker 480

Speaker 481

Speaker 482

Speaker 483

Speaker 484

Speaker 485

Speaker 486

Speaker 487

Speaker 488

Speaker 489

Speaker 490

Speaker 491

Speaker 492

Speaker 493

Speaker 494

Speaker 495

Speaker 496

Speaker 497

Speaker 498

Speaker 499

Speaker 500

Speaker 501

Speaker 502

Speaker 503

Speaker 504

Speaker 505

Speaker 506

Speaker 507

Speaker 508

Speaker 509

Speaker 510

Speaker 511

Speaker 512

Speaker 513

Speaker 514

Speaker 515

Speaker 516

Speaker 517

Speaker 518

Speaker 519

Speaker 520

Speaker 521

Speaker 522

Speaker 523

Speaker 524

Speaker 525

Speaker 526

Speaker 527

Speaker 528

Speaker 529

Speaker 530

Speaker 531

Speaker 532

Speaker 533

Speaker 534

Speaker 535

Speaker 536

Speaker 537

Speaker 538

Speaker 539

Speaker 540

Speaker 541

Speaker 542

Speaker 543

Speaker 544

Speaker 545

Speaker 546

Speaker 547

Speaker 548

Speaker 549

Speaker 550

Speaker 551

Speaker 552

Speaker 553

Speaker 554

Speaker 555

Speaker 556

Speaker 557

Speaker 558

Speaker 559

Speaker 560

Speaker 561

Speaker 562

Speaker 563

Speaker 564

Speaker 565

Speaker 566

Speaker 567

Speaker 568

Speaker 569

Speaker 570

Speaker 571

Speaker 572

Speaker 573

Speaker 574

Speaker 575

Speaker 576

Speaker 577

Speaker 578

Speaker 579

Speaker 580

Speaker 581

Speaker 582

Speaker 583

Speaker 584

Speaker 585

Speaker 586

Speaker 587

Speaker 588

Speaker 589

Speaker 590

Speaker 591

Speaker 592

Speaker 593

Speaker 594

Speaker 595

Speaker 596

Speaker 597

Speaker 598

Speaker 599

Speaker 600

Speaker 601

Speaker 602

Speaker 603

Speaker 604

Speaker 605

Speaker 606

Speaker 607

Speaker 608

Speaker 609

Speaker 610

Speaker 611

Speaker 612

Speaker 613

Speaker 614

Speaker 615

Speaker 616

Speaker 617

Speaker 618

Speaker 619

Speaker 620

Speaker 621

Speaker 622

Speaker 623

Speaker 624

Speaker 625

Speaker 626

Speaker 627

Speaker 628

Speaker 629

Speaker 630

Speaker 631

Speaker 632

Speaker 633

Speaker 634

Speaker 635

Speaker 636

Speaker 637

Speaker 638

Speaker 639

Speaker 640

Speaker 641

Speaker 642

Speaker 643

Speaker 644

Speaker 645

Speaker 646

Speaker 647

Speaker 648

Speaker 649

Speaker 650

Speaker 651

Speaker 652

Speaker 653

Speaker 654

Speaker 655

Speaker 656

Speaker 657

Speaker 658

Speaker 659

Speaker 660

Speaker 661

Speaker 662

Speaker 663

Speaker 664

Speaker 665

Speaker 666

Speaker 667

Speaker 668

Speaker 669

Speaker 670

Speaker 671

Speaker 672

Speaker 673

Speaker 674

Speaker 675

Speaker 676

Speaker 677

Speaker 678

Speaker 679

Speaker 680

Speaker 681

Speaker 682

Speaker 683

Speaker 684

Speaker 685

Speaker 686

Speaker 687

Speaker 688

Speaker 689

Speaker 690

Speaker 691

Speaker 692

Speaker 693

Speaker 694

Speaker 695

Speaker 696

Speaker 697

Speaker 698

Speaker 699

Speaker 700

Speaker 701

Speaker 702

Speaker 703

Speaker 704

Speaker 705

Speaker 706

Speaker 707

Speaker 708

Speaker 709

Speaker 710

Speaker 711

Speaker 712

Speaker 713

Speaker 714

Speaker 715

Speaker 716

Speaker 717

Speaker 718

Speaker 719

Speaker 720

Speaker 721

Speaker 722

Speaker 723

Speaker 724

Speaker 725

Speaker 726

Speaker 727

Speaker 728

Speaker 729

Speaker 730

Speaker 731

Speaker 732

Speaker 733

Speaker 734

Speaker 735

Speaker 736

Speaker 737

Speaker 738

Speaker 739

Speaker 740

Speaker 741

Speaker 742

Speaker 743

Speaker 744

Speaker 745

Speaker 746

Speaker 747

Speaker 748

Speaker 749

Speaker 750

Speaker 751

Speaker 752

Speaker 753

Speaker 754

Speaker 755

Speaker 756

Speaker 757

Speaker 758

Speaker 759

Speaker 760

Speaker 761

Speaker 762

Speaker 763

Speaker 764

Speaker 765

Speaker 766

Speaker 767

Speaker 768

Speaker 769

Speaker 770

Speaker 771

Speaker 772

Speaker 773

Speaker 774

Speaker 775

Speaker 776

Speaker 777

Speaker 778

Speaker 779

Speaker 780

Speaker 781

Speaker 782

Speaker 783

Speaker 784

Speaker 785

Speaker 786

Speaker 787

Speaker 788

Speaker 789

Speaker 790

Speaker 791

Speaker 792

Speaker 793

Speaker 794

Speaker 795

Speaker 796

Speaker 797

Speaker 798

Speaker 799

Speaker 800

Speaker 801

Speaker 802

Speaker 803

Speaker 804

Speaker 805

Speaker 806

Speaker 807

Speaker 808

Speaker 809

Speaker 810

Speaker 811

Speaker 812

Speaker 813

Speaker 814

Speaker 815

Speaker 816

Speaker 817

Speaker 818

Speaker 819

Speaker 820

Speaker 821

Speaker 822

Speaker 823

Speaker 824

Speaker 825

Speaker 826

Speaker 827

Speaker 828

Speaker 829

Speaker 830

Speaker 831

Speaker 832

Speaker 833

Speaker 834

Speaker 835

Speaker 836

Speaker 837

Speaker 838

Speaker 839

Speaker 840

Speaker 841

Speaker 842

Speaker 843

Speaker 844

Speaker 845

Speaker 846

Speaker 847

Speaker 848

Speaker 849

Speaker 850

Speaker 851

Speaker 852

Speaker 853

Speaker 854

Speaker 855

Speaker 856

Speaker 857

Speaker 858

Speaker 859

Speaker 860

Speaker 861

Speaker 862

Speaker 863

Speaker 864

Speaker 865

Speaker 866

Speaker 867

Speaker 868

Speaker 869

Speaker 870

Speaker 871

Speaker 872

Speaker 873

Speaker 874

Speaker 875

Speaker 876

Speaker 877

Speaker 878

Speaker 879

Speaker 880

Speaker 881

Speaker 882

Speaker 883

Speaker 884

Speaker 885

Speaker 886

Speaker 887

Speaker 888

Speaker 889

Speaker 890

Speaker 891

Speaker 892

Speaker 893

Speaker 894

Speaker 895

Speaker 896

Speaker 897

Speaker 898

Speaker 899

Speaker 900

Speaker 901

Speaker 902

Speaker 903

vits-piper-en_US-ljspeech-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/ljspeech/high

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ljspeech-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-ljspeech-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-ljspeech-high/en_US-ljspeech-high.onnx";
  config.model.vits.tokens = "vits-piper-en_US-ljspeech-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-ljspeech-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ljspeech-high.tar.bz2

You can use the following code to play with vits-piper-en_US-ljspeech-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-ljspeech-high/en_US-ljspeech-high.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-ljspeech-high/espeak-ng-data",
            tokens="vits-piper-en_US-ljspeech-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-ljspeech-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/ljspeech/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ljspeech-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-ljspeech-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-ljspeech-medium/en_US-ljspeech-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-ljspeech-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-ljspeech-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ljspeech-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-ljspeech-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-ljspeech-medium/en_US-ljspeech-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-ljspeech-medium/espeak-ng-data",
            tokens="vits-piper-en_US-ljspeech-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-norman-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/norman/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-norman-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-norman-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-norman-medium/en_US-norman-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-norman-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-norman-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-norman-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-norman-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-norman-medium/en_US-norman-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-norman-medium/espeak-ng-data",
            tokens="vits-piper-en_US-norman-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-reza_ibrahim-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/reza_ibrahim/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-reza_ibrahim-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-reza_ibrahim-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-reza_ibrahim-medium/en_US-reza_ibrahim-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-reza_ibrahim-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-reza_ibrahim-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-reza_ibrahim-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-reza_ibrahim-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-reza_ibrahim-medium/en_US-reza_ibrahim-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-reza_ibrahim-medium/espeak-ng-data",
            tokens="vits-piper-en_US-reza_ibrahim-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-ryan-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/ryan/high

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ryan-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-ryan-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-ryan-high/en_US-ryan-high.onnx";
  config.model.vits.tokens = "vits-piper-en_US-ryan-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-ryan-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ryan-high.tar.bz2

You can use the following code to play with vits-piper-en_US-ryan-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-ryan-high/en_US-ryan-high.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-ryan-high/espeak-ng-data",
            tokens="vits-piper-en_US-ryan-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-ryan-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/ryan/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ryan-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-ryan-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-ryan-low/en_US-ryan-low.onnx";
  config.model.vits.tokens = "vits-piper-en_US-ryan-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-ryan-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ryan-low.tar.bz2

You can use the following code to play with vits-piper-en_US-ryan-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-ryan-low/en_US-ryan-low.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-ryan-low/espeak-ng-data",
            tokens="vits-piper-en_US-ryan-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-ryan-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/ryan/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ryan-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-ryan-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-ryan-medium/en_US-ryan-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-ryan-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-ryan-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-ryan-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-ryan-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-ryan-medium/en_US-ryan-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-ryan-medium/espeak-ng-data",
            tokens="vits-piper-en_US-ryan-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-en_US-sam-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/en/en_US/sam/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-sam-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-en_US-sam-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-en_US-sam-medium/en_US-sam-medium.onnx";
  config.model.vits.tokens = "vits-piper-en_US-sam-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-en_US-sam-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-sam-medium.tar.bz2

You can use the following code to play with vits-piper-en_US-sam-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-en_US-sam-medium/en_US-sam-medium.onnx",
            lexicon="",
            data_dir="vits-piper-en_US-sam-medium/espeak-ng-data",
            tokens="vits-piper-en_US-sam-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Friends fell out often because life was changing so fast.
The easiest thing in the world was to lose touch with someone.

sample audios for different speakers are listed below:

Speaker 0

Finnish

This section lists text to speech models for Finnish.

vits-piper-fi_FI-harri-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fi/fi_FI/harri/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fi_FI-harri-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fi_FI-harri-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fi_FI-harri-low/fi_FI-harri-low.onnx";
  config.model.vits.tokens = "vits-piper-fi_FI-harri-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fi_FI-harri-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Sateenkaaren päässä on kultaa, mutta vain ne, jotka siihen uskovat, voivat sen löytää.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fi_FI-harri-low.tar.bz2

You can use the following code to play with vits-piper-fi_FI-harri-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fi_FI-harri-low/fi_FI-harri-low.onnx",
            lexicon="",
            data_dir="vits-piper-fi_FI-harri-low/espeak-ng-data",
            tokens="vits-piper-fi_FI-harri-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Sateenkaaren päässä on kultaa, mutta vain ne, jotka siihen uskovat, voivat sen löytää.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Sateenkaaren päässä on kultaa, mutta vain ne, jotka siihen uskovat, voivat sen löytää.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fi_FI-harri-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fi/fi_FI/harri/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fi_FI-harri-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fi_FI-harri-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fi_FI-harri-medium/fi_FI-harri-medium.onnx";
  config.model.vits.tokens = "vits-piper-fi_FI-harri-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fi_FI-harri-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Sateenkaaren päässä on kultaa, mutta vain ne, jotka siihen uskovat, voivat sen löytää.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fi_FI-harri-medium.tar.bz2

You can use the following code to play with vits-piper-fi_FI-harri-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fi_FI-harri-medium/fi_FI-harri-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fi_FI-harri-medium/espeak-ng-data",
            tokens="vits-piper-fi_FI-harri-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Sateenkaaren päässä on kultaa, mutta vain ne, jotka siihen uskovat, voivat sen löytää.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Sateenkaaren päässä on kultaa, mutta vain ne, jotka siihen uskovat, voivat sen löytää.

sample audios for different speakers are listed below:

Speaker 0

French

This section lists text to speech models for French.

vits-piper-fr_FR-gilles-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fr/fr_FR/gilles/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-gilles-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fr_FR-gilles-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fr_FR-gilles-low/fr_FR-gilles-low.onnx";
  config.model.vits.tokens = "vits-piper-fr_FR-gilles-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fr_FR-gilles-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Pas de nouvelles, bonnes nouvelles.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-gilles-low.tar.bz2

You can use the following code to play with vits-piper-fr_FR-gilles-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fr_FR-gilles-low/fr_FR-gilles-low.onnx",
            lexicon="",
            data_dir="vits-piper-fr_FR-gilles-low/espeak-ng-data",
            tokens="vits-piper-fr_FR-gilles-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Pas de nouvelles, bonnes nouvelles.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Pas de nouvelles, bonnes nouvelles.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fr_FR-siwis-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fr/fr_FR/siwis/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-siwis-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fr_FR-siwis-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fr_FR-siwis-low/fr_FR-siwis-low.onnx";
  config.model.vits.tokens = "vits-piper-fr_FR-siwis-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fr_FR-siwis-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Pas de nouvelles, bonnes nouvelles.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-siwis-low.tar.bz2

You can use the following code to play with vits-piper-fr_FR-siwis-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fr_FR-siwis-low/fr_FR-siwis-low.onnx",
            lexicon="",
            data_dir="vits-piper-fr_FR-siwis-low/espeak-ng-data",
            tokens="vits-piper-fr_FR-siwis-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Pas de nouvelles, bonnes nouvelles.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Pas de nouvelles, bonnes nouvelles.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fr_FR-siwis-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fr/fr_FR/siwis/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-siwis-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fr_FR-siwis-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fr_FR-siwis-medium/fr_FR-siwis-medium.onnx";
  config.model.vits.tokens = "vits-piper-fr_FR-siwis-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fr_FR-siwis-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Pas de nouvelles, bonnes nouvelles.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-siwis-medium.tar.bz2

You can use the following code to play with vits-piper-fr_FR-siwis-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fr_FR-siwis-medium/fr_FR-siwis-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fr_FR-siwis-medium/espeak-ng-data",
            tokens="vits-piper-fr_FR-siwis-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Pas de nouvelles, bonnes nouvelles.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Pas de nouvelles, bonnes nouvelles.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fr_FR-tom-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fr/fr_FR/tom/medium

Number of speakers	Sample rate
1	44100

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-tom-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fr_FR-tom-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fr_FR-tom-medium/fr_FR-tom-medium.onnx";
  config.model.vits.tokens = "vits-piper-fr_FR-tom-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fr_FR-tom-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Pas de nouvelles, bonnes nouvelles.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-tom-medium.tar.bz2

You can use the following code to play with vits-piper-fr_FR-tom-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fr_FR-tom-medium/fr_FR-tom-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fr_FR-tom-medium/espeak-ng-data",
            tokens="vits-piper-fr_FR-tom-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Pas de nouvelles, bonnes nouvelles.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Pas de nouvelles, bonnes nouvelles.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fr_FR-upmc-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fr/fr_FR/upmc/medium

Number of speakers	Sample rate
2	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-upmc-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fr_FR-upmc-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fr_FR-upmc-medium/fr_FR-upmc-medium.onnx";
  config.model.vits.tokens = "vits-piper-fr_FR-upmc-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fr_FR-upmc-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Pas de nouvelles, bonnes nouvelles.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fr_FR-upmc-medium.tar.bz2

You can use the following code to play with vits-piper-fr_FR-upmc-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fr_FR-upmc-medium/fr_FR-upmc-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fr_FR-upmc-medium/espeak-ng-data",
            tokens="vits-piper-fr_FR-upmc-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Pas de nouvelles, bonnes nouvelles.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Pas de nouvelles, bonnes nouvelles.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Georgian

This section lists text to speech models for Georgian.

vits-piper-ka_GE-natia-medium

vits-piper-ka_GE-natia-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ka/ka_GE/natia/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ka_GE-natia-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ka_GE-natia-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ka_GE-natia-medium/ka_GE-natia-medium.onnx";
  config.model.vits.tokens = "vits-piper-ka_GE-natia-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ka_GE-natia-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "ღვინო თბილისში, საქართველო სამტრედში";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ka_GE-natia-medium.tar.bz2

You can use the following code to play with vits-piper-ka_GE-natia-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ka_GE-natia-medium/ka_GE-natia-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ka_GE-natia-medium/espeak-ng-data",
            tokens="vits-piper-ka_GE-natia-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="ღვინო თბილისში, საქართველო სამტრედში",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

ღვინო თბილისში, საქართველო სამტრედში

sample audios for different speakers are listed below:

Speaker 0

German

This section lists text to speech models for German.

vits-piper-de_DE-eva_k-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/eva_k/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-eva_k-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-eva_k-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-eva_k-x_low/de_DE-eva_k-x_low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-eva_k-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-eva_k-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-eva_k-x_low.tar.bz2

You can use the following code to play with vits-piper-de_DE-eva_k-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-eva_k-x_low/de_DE-eva_k-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-eva_k-x_low/espeak-ng-data",
            tokens="vits-piper-de_DE-eva_k-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-glados-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/systemofapwne/piper-de-glados

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-glados-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-glados-high/de_DE-glados-high.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-glados-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-glados-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados-high.tar.bz2

You can use the following code to play with vits-piper-de_DE-glados-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-glados-high/de_DE-glados-high.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-glados-high/espeak-ng-data",
            tokens="vits-piper-de_DE-glados-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-glados-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/systemofapwne/piper-de-glados

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-glados-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-glados-low/de_DE-glados-low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-glados-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-glados-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados-low.tar.bz2

You can use the following code to play with vits-piper-de_DE-glados-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-glados-low/de_DE-glados-low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-glados-low/espeak-ng-data",
            tokens="vits-piper-de_DE-glados-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-glados-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/systemofapwne/piper-de-glados

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-glados-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-glados-medium/de_DE-glados-medium.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-glados-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-glados-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados-medium.tar.bz2

You can use the following code to play with vits-piper-de_DE-glados-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-glados-medium/de_DE-glados-medium.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-glados-medium/espeak-ng-data",
            tokens="vits-piper-de_DE-glados-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-glados_turret-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/systemofapwne/piper-de-glados

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados_turret-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-glados_turret-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-glados_turret-high/de_DE-glados_turret-high.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-glados_turret-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-glados_turret-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados_turret-high.tar.bz2

You can use the following code to play with vits-piper-de_DE-glados_turret-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-glados_turret-high/de_DE-glados_turret-high.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-glados_turret-high/espeak-ng-data",
            tokens="vits-piper-de_DE-glados_turret-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-glados_turret-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/systemofapwne/piper-de-glados

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados_turret-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-glados_turret-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-glados_turret-low/de_DE-glados_turret-low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-glados_turret-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-glados_turret-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados_turret-low.tar.bz2

You can use the following code to play with vits-piper-de_DE-glados_turret-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-glados_turret-low/de_DE-glados_turret-low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-glados_turret-low/espeak-ng-data",
            tokens="vits-piper-de_DE-glados_turret-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-glados_turret-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/systemofapwne/piper-de-glados

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados_turret-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-glados_turret-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-glados_turret-medium/de_DE-glados_turret-medium.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-glados_turret-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-glados_turret-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-glados_turret-medium.tar.bz2

You can use the following code to play with vits-piper-de_DE-glados_turret-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-glados_turret-medium/de_DE-glados_turret-medium.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-glados_turret-medium/espeak-ng-data",
            tokens="vits-piper-de_DE-glados_turret-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-karlsson-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/karlsson/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-karlsson-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-karlsson-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-karlsson-low/de_DE-karlsson-low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-karlsson-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-karlsson-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-karlsson-low.tar.bz2

You can use the following code to play with vits-piper-de_DE-karlsson-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-karlsson-low/de_DE-karlsson-low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-karlsson-low/espeak-ng-data",
            tokens="vits-piper-de_DE-karlsson-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-kerstin-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/kerstin/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-kerstin-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-kerstin-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-kerstin-low/de_DE-kerstin-low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-kerstin-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-kerstin-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-kerstin-low.tar.bz2

You can use the following code to play with vits-piper-de_DE-kerstin-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-kerstin-low/de_DE-kerstin-low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-kerstin-low/espeak-ng-data",
            tokens="vits-piper-de_DE-kerstin-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-pavoque-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/pavoque/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-pavoque-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-pavoque-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-pavoque-low/de_DE-pavoque-low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-pavoque-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-pavoque-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-pavoque-low.tar.bz2

You can use the following code to play with vits-piper-de_DE-pavoque-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-pavoque-low/de_DE-pavoque-low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-pavoque-low/espeak-ng-data",
            tokens="vits-piper-de_DE-pavoque-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-ramona-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/ramona/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-ramona-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-ramona-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-ramona-low/de_DE-ramona-low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-ramona-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-ramona-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-ramona-low.tar.bz2

You can use the following code to play with vits-piper-de_DE-ramona-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-ramona-low/de_DE-ramona-low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-ramona-low/espeak-ng-data",
            tokens="vits-piper-de_DE-ramona-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-thorsten-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/thorsten/high

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-thorsten-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-thorsten-high/de_DE-thorsten-high.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-thorsten-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-thorsten-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten-high.tar.bz2

You can use the following code to play with vits-piper-de_DE-thorsten-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-thorsten-high/de_DE-thorsten-high.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-thorsten-high/espeak-ng-data",
            tokens="vits-piper-de_DE-thorsten-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-thorsten-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/thorsten/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-thorsten-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-thorsten-low/de_DE-thorsten-low.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-thorsten-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-thorsten-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten-low.tar.bz2

You can use the following code to play with vits-piper-de_DE-thorsten-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-thorsten-low/de_DE-thorsten-low.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-thorsten-low/espeak-ng-data",
            tokens="vits-piper-de_DE-thorsten-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-thorsten-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/thorsten/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-thorsten-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-thorsten-medium/de_DE-thorsten-medium.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-thorsten-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-thorsten-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten-medium.tar.bz2

You can use the following code to play with vits-piper-de_DE-thorsten-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-thorsten-medium/de_DE-thorsten-medium.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-thorsten-medium/espeak-ng-data",
            tokens="vits-piper-de_DE-thorsten-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-de_DE-thorsten_emotional-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/de/de_DE/thorsten_emotional/medium

Number of speakers	Sample rate
8	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten_emotional-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-de_DE-thorsten_emotional-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-de_DE-thorsten_emotional-medium/de_DE-thorsten_emotional-medium.onnx";
  config.model.vits.tokens = "vits-piper-de_DE-thorsten_emotional-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-de_DE-thorsten_emotional-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Alles hat ein Ende, nur die Wurst hat zwei.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-de_DE-thorsten_emotional-medium.tar.bz2

You can use the following code to play with vits-piper-de_DE-thorsten_emotional-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-de_DE-thorsten_emotional-medium/de_DE-thorsten_emotional-medium.onnx",
            lexicon="",
            data_dir="vits-piper-de_DE-thorsten_emotional-medium/espeak-ng-data",
            tokens="vits-piper-de_DE-thorsten_emotional-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Alles hat ein Ende, nur die Wurst hat zwei.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Alles hat ein Ende, nur die Wurst hat zwei.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Greek

This section lists text to speech models for Greek.

vits-piper-el_GR-rapunzelina-low

vits-piper-el_GR-rapunzelina-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/el/el_GR/rapunzelina/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-el_GR-rapunzelina-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-el_GR-rapunzelina-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-el_GR-rapunzelina-low/el_GR-rapunzelina-low.onnx";
  config.model.vits.tokens = "vits-piper-el_GR-rapunzelina-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-el_GR-rapunzelina-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Όταν το δέντρο είναι μικρό, το στρέβλεις· όταν είναι μεγάλο, το λυγίζεις.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-el_GR-rapunzelina-low.tar.bz2

You can use the following code to play with vits-piper-el_GR-rapunzelina-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-el_GR-rapunzelina-low/el_GR-rapunzelina-low.onnx",
            lexicon="",
            data_dir="vits-piper-el_GR-rapunzelina-low/espeak-ng-data",
            tokens="vits-piper-el_GR-rapunzelina-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Όταν το δέντρο είναι μικρό, το στρέβλεις· όταν είναι μεγάλο, το λυγίζεις.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Όταν το δέντρο είναι μικρό, το στρέβλεις· όταν είναι μεγάλο, το λυγίζεις.

sample audios for different speakers are listed below:

Speaker 0

Hungarian

This section lists text to speech models for Hungarian.

vits-piper-hu_HU-anna-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/hu/hu_HU/anna/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-hu_HU-anna-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-hu_HU-anna-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-hu_HU-anna-medium/hu_HU-anna-medium.onnx";
  config.model.vits.tokens = "vits-piper-hu_HU-anna-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-hu_HU-anna-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Ha északról fúj a szél, a lányok nem lógnak együtt.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-hu_HU-anna-medium.tar.bz2

You can use the following code to play with vits-piper-hu_HU-anna-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-hu_HU-anna-medium/hu_HU-anna-medium.onnx",
            lexicon="",
            data_dir="vits-piper-hu_HU-anna-medium/espeak-ng-data",
            tokens="vits-piper-hu_HU-anna-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Ha északról fúj a szél, a lányok nem lógnak együtt.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Ha északról fúj a szél, a lányok nem lógnak együtt.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-hu_HU-berta-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/hu/hu_HU/berta/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-hu_HU-berta-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-hu_HU-berta-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-hu_HU-berta-medium/hu_HU-berta-medium.onnx";
  config.model.vits.tokens = "vits-piper-hu_HU-berta-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-hu_HU-berta-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Ha északról fúj a szél, a lányok nem lógnak együtt.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-hu_HU-berta-medium.tar.bz2

You can use the following code to play with vits-piper-hu_HU-berta-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-hu_HU-berta-medium/hu_HU-berta-medium.onnx",
            lexicon="",
            data_dir="vits-piper-hu_HU-berta-medium/espeak-ng-data",
            tokens="vits-piper-hu_HU-berta-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Ha északról fúj a szél, a lányok nem lógnak együtt.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Ha északról fúj a szél, a lányok nem lógnak együtt.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-hu_HU-imre-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/hu/hu_HU/imre/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-hu_HU-imre-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-hu_HU-imre-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-hu_HU-imre-medium/hu_HU-imre-medium.onnx";
  config.model.vits.tokens = "vits-piper-hu_HU-imre-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-hu_HU-imre-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Ha északról fúj a szél, a lányok nem lógnak együtt.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-hu_HU-imre-medium.tar.bz2

You can use the following code to play with vits-piper-hu_HU-imre-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-hu_HU-imre-medium/hu_HU-imre-medium.onnx",
            lexicon="",
            data_dir="vits-piper-hu_HU-imre-medium/espeak-ng-data",
            tokens="vits-piper-hu_HU-imre-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Ha északról fúj a szél, a lányok nem lógnak együtt.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Ha északról fúj a szél, a lányok nem lógnak együtt.

sample audios for different speakers are listed below:

Speaker 0

Icelandic

This section lists text to speech models for Icelandic.

vits-piper-is_IS-bui-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/is/is_IS/bui/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-bui-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-is_IS-bui-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-is_IS-bui-medium/is_IS-bui-medium.onnx";
  config.model.vits.tokens = "vits-piper-is_IS-bui-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-is_IS-bui-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Farðu með allt, eða farðu ekki.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-bui-medium.tar.bz2

You can use the following code to play with vits-piper-is_IS-bui-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-is_IS-bui-medium/is_IS-bui-medium.onnx",
            lexicon="",
            data_dir="vits-piper-is_IS-bui-medium/espeak-ng-data",
            tokens="vits-piper-is_IS-bui-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Farðu með allt, eða farðu ekki.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Farðu með allt, eða farðu ekki.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-is_IS-salka-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/is/is_IS/salka/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-salka-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-is_IS-salka-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-is_IS-salka-medium/is_IS-salka-medium.onnx";
  config.model.vits.tokens = "vits-piper-is_IS-salka-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-is_IS-salka-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Farðu með allt, eða farðu ekki.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-salka-medium.tar.bz2

You can use the following code to play with vits-piper-is_IS-salka-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-is_IS-salka-medium/is_IS-salka-medium.onnx",
            lexicon="",
            data_dir="vits-piper-is_IS-salka-medium/espeak-ng-data",
            tokens="vits-piper-is_IS-salka-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Farðu með allt, eða farðu ekki.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Farðu með allt, eða farðu ekki.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-is_IS-steinn-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/is/is_IS/steinn/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-steinn-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-is_IS-steinn-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-is_IS-steinn-medium/is_IS-steinn-medium.onnx";
  config.model.vits.tokens = "vits-piper-is_IS-steinn-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-is_IS-steinn-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Farðu með allt, eða farðu ekki.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-steinn-medium.tar.bz2

You can use the following code to play with vits-piper-is_IS-steinn-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-is_IS-steinn-medium/is_IS-steinn-medium.onnx",
            lexicon="",
            data_dir="vits-piper-is_IS-steinn-medium/espeak-ng-data",
            tokens="vits-piper-is_IS-steinn-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Farðu með allt, eða farðu ekki.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Farðu með allt, eða farðu ekki.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-is_IS-ugla-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/is/is_IS/ugla/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-ugla-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-is_IS-ugla-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-is_IS-ugla-medium/is_IS-ugla-medium.onnx";
  config.model.vits.tokens = "vits-piper-is_IS-ugla-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-is_IS-ugla-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Farðu með allt, eða farðu ekki.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-is_IS-ugla-medium.tar.bz2

You can use the following code to play with vits-piper-is_IS-ugla-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-is_IS-ugla-medium/is_IS-ugla-medium.onnx",
            lexicon="",
            data_dir="vits-piper-is_IS-ugla-medium/espeak-ng-data",
            tokens="vits-piper-is_IS-ugla-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Farðu með allt, eða farðu ekki.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Farðu með allt, eða farðu ekki.

sample audios for different speakers are listed below:

Speaker 0

Italian

This section lists text to speech models for Italian.

vits-piper-it_IT-paola-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/it/it_IT/paola/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-it_IT-paola-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-it_IT-paola-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-it_IT-paola-medium/it_IT-paola-medium.onnx";
  config.model.vits.tokens = "vits-piper-it_IT-paola-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-it_IT-paola-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Se vuoi andare veloce, vai da solo; se vuoi andare lontano, vai insieme.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-it_IT-paola-medium.tar.bz2

You can use the following code to play with vits-piper-it_IT-paola-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-it_IT-paola-medium/it_IT-paola-medium.onnx",
            lexicon="",
            data_dir="vits-piper-it_IT-paola-medium/espeak-ng-data",
            tokens="vits-piper-it_IT-paola-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Se vuoi andare veloce, vai da solo; se vuoi andare lontano, vai insieme.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Se vuoi andare veloce, vai da solo; se vuoi andare lontano, vai insieme.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-it_IT-riccardo-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/it/it_IT/riccardo/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-it_IT-riccardo-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-it_IT-riccardo-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-it_IT-riccardo-x_low/it_IT-riccardo-x_low.onnx";
  config.model.vits.tokens = "vits-piper-it_IT-riccardo-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-it_IT-riccardo-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Se vuoi andare veloce, vai da solo; se vuoi andare lontano, vai insieme.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-it_IT-riccardo-x_low.tar.bz2

You can use the following code to play with vits-piper-it_IT-riccardo-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-it_IT-riccardo-x_low/it_IT-riccardo-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-it_IT-riccardo-x_low/espeak-ng-data",
            tokens="vits-piper-it_IT-riccardo-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Se vuoi andare veloce, vai da solo; se vuoi andare lontano, vai insieme.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Se vuoi andare veloce, vai da solo; se vuoi andare lontano, vai insieme.

sample audios for different speakers are listed below:

Speaker 0

Kazakh

This section lists text to speech models for Kazakh.

vits-piper-kk_KZ-iseke-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/kk/kk_KZ/iseke/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-kk_KZ-iseke-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-kk_KZ-iseke-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-kk_KZ-iseke-x_low/kk_KZ-iseke-x_low.onnx";
  config.model.vits.tokens = "vits-piper-kk_KZ-iseke-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-kk_KZ-iseke-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Әлемнің жұлдыздары сенің көзің, жаным.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-kk_KZ-iseke-x_low.tar.bz2

You can use the following code to play with vits-piper-kk_KZ-iseke-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-kk_KZ-iseke-x_low/kk_KZ-iseke-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-kk_KZ-iseke-x_low/espeak-ng-data",
            tokens="vits-piper-kk_KZ-iseke-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Әлемнің жұлдыздары сенің көзің, жаным.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Әлемнің жұлдыздары сенің көзің, жаным.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-kk_KZ-issai-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/kk/kk_KZ/issai/high

Number of speakers	Sample rate
6	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-kk_KZ-issai-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-kk_KZ-issai-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-kk_KZ-issai-high/kk_KZ-issai-high.onnx";
  config.model.vits.tokens = "vits-piper-kk_KZ-issai-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-kk_KZ-issai-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Әлемнің жұлдыздары сенің көзің, жаным.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-kk_KZ-issai-high.tar.bz2

You can use the following code to play with vits-piper-kk_KZ-issai-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-kk_KZ-issai-high/kk_KZ-issai-high.onnx",
            lexicon="",
            data_dir="vits-piper-kk_KZ-issai-high/espeak-ng-data",
            tokens="vits-piper-kk_KZ-issai-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Әлемнің жұлдыздары сенің көзің, жаным.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Әлемнің жұлдыздары сенің көзің, жаным.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

vits-piper-kk_KZ-raya-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/kk/kk_KZ/raya/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-kk_KZ-raya-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-kk_KZ-raya-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-kk_KZ-raya-x_low/kk_KZ-raya-x_low.onnx";
  config.model.vits.tokens = "vits-piper-kk_KZ-raya-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-kk_KZ-raya-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Әлемнің жұлдыздары сенің көзің, жаным.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-kk_KZ-raya-x_low.tar.bz2

You can use the following code to play with vits-piper-kk_KZ-raya-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-kk_KZ-raya-x_low/kk_KZ-raya-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-kk_KZ-raya-x_low/espeak-ng-data",
            tokens="vits-piper-kk_KZ-raya-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Әлемнің жұлдыздары сенің көзің, жаным.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Әлемнің жұлдыздары сенің көзің, жаным.

sample audios for different speakers are listed below:

Speaker 0

Latvian

This section lists text to speech models for Latvian.

vits-piper-lv_LV-aivars-medium

vits-piper-lv_LV-aivars-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/lv/lv_LV/aivars/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-lv_LV-aivars-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-lv_LV-aivars-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-lv_LV-aivars-medium/lv_LV-aivars-medium.onnx";
  config.model.vits.tokens = "vits-piper-lv_LV-aivars-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-lv_LV-aivars-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Zeme nenes augļus, ja tēvs sēj, bet māte auž.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-lv_LV-aivars-medium.tar.bz2

You can use the following code to play with vits-piper-lv_LV-aivars-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-lv_LV-aivars-medium/lv_LV-aivars-medium.onnx",
            lexicon="",
            data_dir="vits-piper-lv_LV-aivars-medium/espeak-ng-data",
            tokens="vits-piper-lv_LV-aivars-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Zeme nenes augļus, ja tēvs sēj, bet māte auž.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Zeme nenes augļus, ja tēvs sēj, bet māte auž.

sample audios for different speakers are listed below:

Speaker 0

Luxembourgish

This section lists text to speech models for Luxembourgish.

vits-piper-lb_LU-marylux-medium

vits-piper-lb_LU-marylux-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/lb/lb_LU/marylux/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-lb_LU-marylux-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-lb_LU-marylux-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-lb_LU-marylux-medium/lb_LU-marylux-medium.onnx";
  config.model.vits.tokens = "vits-piper-lb_LU-marylux-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-lb_LU-marylux-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Op der Haaptstrooss sinn all Stroossen Brécken, awer d'Dier kann iwwerall erreecht ginn.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-lb_LU-marylux-medium.tar.bz2

You can use the following code to play with vits-piper-lb_LU-marylux-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-lb_LU-marylux-medium/lb_LU-marylux-medium.onnx",
            lexicon="",
            data_dir="vits-piper-lb_LU-marylux-medium/espeak-ng-data",
            tokens="vits-piper-lb_LU-marylux-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Op der Haaptstrooss sinn all Stroossen Brécken, awer d'Dier kann iwwerall erreecht ginn.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Op der Haaptstrooss sinn all Stroossen Brécken, awer d'Dier kann iwwerall erreecht ginn.

sample audios for different speakers are listed below:

Speaker 0

Malayalam

This section lists text to speech models for Malayalam.

vits-piper-ml_IN-arjun-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ml/ml_IN/arjun/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ml_IN-arjun-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ml_IN-arjun-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ml_IN-arjun-medium/ml_IN-arjun-medium.onnx";
  config.model.vits.tokens = "vits-piper-ml_IN-arjun-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ml_IN-arjun-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "മണ്ണ് മരിക്കുമ്പോൾ കാട്ടിലെ വെള്ളവും മരിക്കുന്നു.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ml_IN-arjun-medium.tar.bz2

You can use the following code to play with vits-piper-ml_IN-arjun-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ml_IN-arjun-medium/ml_IN-arjun-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ml_IN-arjun-medium/espeak-ng-data",
            tokens="vits-piper-ml_IN-arjun-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="മണ്ണ് മരിക്കുമ്പോൾ കാട്ടിലെ വെള്ളവും മരിക്കുന്നു.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

മണ്ണ് മരിക്കുമ്പോൾ കാട്ടിലെ വെള്ളവും മരിക്കുന്നു.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ml_IN-meera-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ml/ml_IN/meera/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ml_IN-meera-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ml_IN-meera-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ml_IN-meera-medium/ml_IN-meera-medium.onnx";
  config.model.vits.tokens = "vits-piper-ml_IN-meera-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ml_IN-meera-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "മണ്ണ് മരിക്കുമ്പോൾ കാട്ടിലെ വെള്ളവും മരിക്കുന്നു.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ml_IN-meera-medium.tar.bz2

You can use the following code to play with vits-piper-ml_IN-meera-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ml_IN-meera-medium/ml_IN-meera-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ml_IN-meera-medium/espeak-ng-data",
            tokens="vits-piper-ml_IN-meera-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="മണ്ണ് മരിക്കുമ്പോൾ കാട്ടിലെ വെള്ളവും മരിക്കുന്നു.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

മണ്ണ് മരിക്കുമ്പോൾ കാട്ടിലെ വെള്ളവും മരിക്കുന്നു.

sample audios for different speakers are listed below:

Speaker 0

Nepali

This section lists text to speech models for Nepali.

vits-piper-ne_NP-chitwan-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ne/ne_NP/chitwan/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ne_NP-chitwan-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ne_NP-chitwan-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ne_NP-chitwan-medium/ne_NP-chitwan-medium.onnx";
  config.model.vits.tokens = "vits-piper-ne_NP-chitwan-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ne_NP-chitwan-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ne_NP-chitwan-medium.tar.bz2

You can use the following code to play with vits-piper-ne_NP-chitwan-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ne_NP-chitwan-medium/ne_NP-chitwan-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ne_NP-chitwan-medium/espeak-ng-data",
            tokens="vits-piper-ne_NP-chitwan-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ne_NP-google-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ne/ne_NP/google/medium

Number of speakers	Sample rate
18	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ne_NP-google-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ne_NP-google-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ne_NP-google-medium/ne_NP-google-medium.onnx";
  config.model.vits.tokens = "vits-piper-ne_NP-google-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ne_NP-google-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ne_NP-google-medium.tar.bz2

You can use the following code to play with vits-piper-ne_NP-google-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ne_NP-google-medium/ne_NP-google-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ne_NP-google-medium/espeak-ng-data",
            tokens="vits-piper-ne_NP-google-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

vits-piper-ne_NP-google-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ne/ne_NP/google/x_low

Number of speakers	Sample rate
18	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ne_NP-google-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ne_NP-google-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ne_NP-google-x_low/ne_NP-google-x_low.onnx";
  config.model.vits.tokens = "vits-piper-ne_NP-google-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ne_NP-google-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ne_NP-google-x_low.tar.bz2

You can use the following code to play with vits-piper-ne_NP-google-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ne_NP-google-x_low/ne_NP-google-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-ne_NP-google-x_low/espeak-ng-data",
            tokens="vits-piper-ne_NP-google-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

घाँसको पातले पहाडलाई अभिवादन गर्दै झुक्छ।

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

Norwegian

This section lists text to speech models for Norwegian.

vits-piper-no_NO-talesyntese-medium

vits-piper-no_NO-talesyntese-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/no/no_NO/talesyntese/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-no_NO-talesyntese-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-no_NO-talesyntese-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-no_NO-talesyntese-medium/no_NO-talesyntese-medium.onnx";
  config.model.vits.tokens = "vits-piper-no_NO-talesyntese-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-no_NO-talesyntese-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Uskyldig kan stormen veroorzaken";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-no_NO-talesyntese-medium.tar.bz2

You can use the following code to play with vits-piper-no_NO-talesyntese-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-no_NO-talesyntese-medium/no_NO-talesyntese-medium.onnx",
            lexicon="",
            data_dir="vits-piper-no_NO-talesyntese-medium/espeak-ng-data",
            tokens="vits-piper-no_NO-talesyntese-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Uskyldig kan stormen veroorzaken",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Uskyldig kan stormen veroorzaken

sample audios for different speakers are listed below:

Speaker 0

Persian

This section lists text to speech models for Persian.

vits-piper-fa_IR-amir-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fa/fa_IR/amir/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-amir-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fa_IR-amir-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fa_IR-amir-medium/fa_IR-amir-medium.onnx";
  config.model.vits.tokens = "vits-piper-fa_IR-amir-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fa_IR-amir-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-amir-medium.tar.bz2

You can use the following code to play with vits-piper-fa_IR-amir-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fa_IR-amir-medium/fa_IR-amir-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fa_IR-amir-medium/espeak-ng-data",
            tokens="vits-piper-fa_IR-amir-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fa_IR-ganji-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fa/fa_IR/ganji/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-ganji-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fa_IR-ganji-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fa_IR-ganji-medium/fa_IR-ganji-medium.onnx";
  config.model.vits.tokens = "vits-piper-fa_IR-ganji-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fa_IR-ganji-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-ganji-medium.tar.bz2

You can use the following code to play with vits-piper-fa_IR-ganji-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fa_IR-ganji-medium/fa_IR-ganji-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fa_IR-ganji-medium/espeak-ng-data",
            tokens="vits-piper-fa_IR-ganji-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fa_IR-ganji_adabi-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fa/fa_IR/ganji_adabi/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-ganji_adabi-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fa_IR-ganji_adabi-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fa_IR-ganji_adabi-medium/fa_IR-ganji_adabi-medium.onnx";
  config.model.vits.tokens = "vits-piper-fa_IR-ganji_adabi-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fa_IR-ganji_adabi-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-ganji_adabi-medium.tar.bz2

You can use the following code to play with vits-piper-fa_IR-ganji_adabi-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fa_IR-ganji_adabi-medium/fa_IR-ganji_adabi-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fa_IR-ganji_adabi-medium/espeak-ng-data",
            tokens="vits-piper-fa_IR-ganji_adabi-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fa_IR-gyro-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fa/fa_IR/gyro/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-gyro-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fa_IR-gyro-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fa_IR-gyro-medium/fa_IR-gyro-medium.onnx";
  config.model.vits.tokens = "vits-piper-fa_IR-gyro-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fa_IR-gyro-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-gyro-medium.tar.bz2

You can use the following code to play with vits-piper-fa_IR-gyro-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fa_IR-gyro-medium/fa_IR-gyro-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fa_IR-gyro-medium/espeak-ng-data",
            tokens="vits-piper-fa_IR-gyro-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-fa_IR-reza_ibrahim-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/fa/fa_IR/reza_ibrahim/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-reza_ibrahim-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-fa_IR-reza_ibrahim-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-fa_IR-reza_ibrahim-medium/fa_IR-reza_ibrahim-medium.onnx";
  config.model.vits.tokens = "vits-piper-fa_IR-reza_ibrahim-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-fa_IR-reza_ibrahim-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-fa_IR-reza_ibrahim-medium.tar.bz2

You can use the following code to play with vits-piper-fa_IR-reza_ibrahim-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-fa_IR-reza_ibrahim-medium/fa_IR-reza_ibrahim-medium.onnx",
            lexicon="",
            data_dir="vits-piper-fa_IR-reza_ibrahim-medium/espeak-ng-data",
            tokens="vits-piper-fa_IR-reza_ibrahim-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

همانطور که کوه ها در برابر باد و باران پایدارند، اما به مرور زمان خرد و پخش می شوند، انسان نیز باید در برابر مشکلات قوی باشد، اما با خرد و خویشتن داری در زندگی به پیش برود.

sample audios for different speakers are listed below:

Speaker 0

Polish

This section lists text to speech models for Polish.

vits-piper-pl_PL-darkman-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pl/pl_PL/darkman/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pl_PL-darkman-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pl_PL-darkman-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pl_PL-darkman-medium/pl_PL-darkman-medium.onnx";
  config.model.vits.tokens = "vits-piper-pl_PL-darkman-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pl_PL-darkman-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Nieważne, za kogo walczysz, i tak popełnisz błąd";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pl_PL-darkman-medium.tar.bz2

You can use the following code to play with vits-piper-pl_PL-darkman-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pl_PL-darkman-medium/pl_PL-darkman-medium.onnx",
            lexicon="",
            data_dir="vits-piper-pl_PL-darkman-medium/espeak-ng-data",
            tokens="vits-piper-pl_PL-darkman-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Nieważne, za kogo walczysz, i tak popełnisz błąd",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Nieważne, za kogo walczysz, i tak popełnisz błąd

sample audios for different speakers are listed below:

Speaker 0

vits-piper-pl_PL-gosia-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pl/pl_PL/gosia/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pl_PL-gosia-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pl_PL-gosia-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pl_PL-gosia-medium/pl_PL-gosia-medium.onnx";
  config.model.vits.tokens = "vits-piper-pl_PL-gosia-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pl_PL-gosia-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Nieważne, za kogo walczysz, i tak popełnisz błąd";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pl_PL-gosia-medium.tar.bz2

You can use the following code to play with vits-piper-pl_PL-gosia-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pl_PL-gosia-medium/pl_PL-gosia-medium.onnx",
            lexicon="",
            data_dir="vits-piper-pl_PL-gosia-medium/espeak-ng-data",
            tokens="vits-piper-pl_PL-gosia-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Nieważne, za kogo walczysz, i tak popełnisz błąd",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Nieważne, za kogo walczysz, i tak popełnisz błąd

sample audios for different speakers are listed below:

Speaker 0

vits-piper-pl_PL-mc_speech-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pl/pl_PL/mc_speech/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pl_PL-mc_speech-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pl_PL-mc_speech-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pl_PL-mc_speech-medium/pl_PL-mc_speech-medium.onnx";
  config.model.vits.tokens = "vits-piper-pl_PL-mc_speech-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pl_PL-mc_speech-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Nieważne, za kogo walczysz, i tak popełnisz błąd";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pl_PL-mc_speech-medium.tar.bz2

You can use the following code to play with vits-piper-pl_PL-mc_speech-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pl_PL-mc_speech-medium/pl_PL-mc_speech-medium.onnx",
            lexicon="",
            data_dir="vits-piper-pl_PL-mc_speech-medium/espeak-ng-data",
            tokens="vits-piper-pl_PL-mc_speech-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Nieważne, za kogo walczysz, i tak popełnisz błąd",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Nieważne, za kogo walczysz, i tak popełnisz błąd

sample audios for different speakers are listed below:

Speaker 0

Portuguese

This section lists text to speech models for Portuguese.

vits-piper-pt_BR-cadu-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pt/pt_BR/cadu/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-cadu-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pt_BR-cadu-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pt_BR-cadu-medium/pt_BR-cadu-medium.onnx";
  config.model.vits.tokens = "vits-piper-pt_BR-cadu-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pt_BR-cadu-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Marinha sem vento, não chega a porto";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-cadu-medium.tar.bz2

You can use the following code to play with vits-piper-pt_BR-cadu-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pt_BR-cadu-medium/pt_BR-cadu-medium.onnx",
            lexicon="",
            data_dir="vits-piper-pt_BR-cadu-medium/espeak-ng-data",
            tokens="vits-piper-pt_BR-cadu-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Marinha sem vento, não chega a porto",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Marinha sem vento, não chega a porto

sample audios for different speakers are listed below:

Speaker 0

vits-piper-pt_BR-edresson-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pt/pt_BR/edresson/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-edresson-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pt_BR-edresson-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pt_BR-edresson-low/pt_BR-edresson-low.onnx";
  config.model.vits.tokens = "vits-piper-pt_BR-edresson-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pt_BR-edresson-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Marinha sem vento, não chega a porto";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-edresson-low.tar.bz2

You can use the following code to play with vits-piper-pt_BR-edresson-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pt_BR-edresson-low/pt_BR-edresson-low.onnx",
            lexicon="",
            data_dir="vits-piper-pt_BR-edresson-low/espeak-ng-data",
            tokens="vits-piper-pt_BR-edresson-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Marinha sem vento, não chega a porto",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Marinha sem vento, não chega a porto

sample audios for different speakers are listed below:

Speaker 0

vits-piper-pt_BR-faber-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pt/pt_BR/faber/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-faber-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pt_BR-faber-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pt_BR-faber-medium/pt_BR-faber-medium.onnx";
  config.model.vits.tokens = "vits-piper-pt_BR-faber-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pt_BR-faber-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Marinha sem vento, não chega a porto";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-faber-medium.tar.bz2

You can use the following code to play with vits-piper-pt_BR-faber-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pt_BR-faber-medium/pt_BR-faber-medium.onnx",
            lexicon="",
            data_dir="vits-piper-pt_BR-faber-medium/espeak-ng-data",
            tokens="vits-piper-pt_BR-faber-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Marinha sem vento, não chega a porto",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Marinha sem vento, não chega a porto

sample audios for different speakers are listed below:

Speaker 0

vits-piper-pt_BR-jeff-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pt/pt_BR/jeff/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-jeff-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pt_BR-jeff-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pt_BR-jeff-medium/pt_BR-jeff-medium.onnx";
  config.model.vits.tokens = "vits-piper-pt_BR-jeff-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pt_BR-jeff-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Marinha sem vento, não chega a porto";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_BR-jeff-medium.tar.bz2

You can use the following code to play with vits-piper-pt_BR-jeff-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pt_BR-jeff-medium/pt_BR-jeff-medium.onnx",
            lexicon="",
            data_dir="vits-piper-pt_BR-jeff-medium/espeak-ng-data",
            tokens="vits-piper-pt_BR-jeff-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Marinha sem vento, não chega a porto",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Marinha sem vento, não chega a porto

sample audios for different speakers are listed below:

Speaker 0

vits-piper-pt_PT-tugão-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/pt/pt_PT/tugão/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_PT-tugão-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-pt_PT-tugão-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-pt_PT-tugão-medium/pt_PT-tugão-medium.onnx";
  config.model.vits.tokens = "vits-piper-pt_PT-tugão-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-pt_PT-tugão-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Marinha sem vento, não chega a porto";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-pt_PT-tugão-medium.tar.bz2

You can use the following code to play with vits-piper-pt_PT-tugão-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-pt_PT-tugão-medium/pt_PT-tugão-medium.onnx",
            lexicon="",
            data_dir="vits-piper-pt_PT-tugão-medium/espeak-ng-data",
            tokens="vits-piper-pt_PT-tugão-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Marinha sem vento, não chega a porto",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Marinha sem vento, não chega a porto

sample audios for different speakers are listed below:

Speaker 0

Romanian

This section lists text to speech models for Romanian.

vits-piper-ro_RO-mihai-medium

vits-piper-ro_RO-mihai-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ro/ro_RO/mihai/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ro_RO-mihai-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ro_RO-mihai-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ro_RO-mihai-medium/ro_RO-mihai-medium.onnx";
  config.model.vits.tokens = "vits-piper-ro_RO-mihai-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ro_RO-mihai-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Un foc fără lemne se stinge, o lume fără poveste moare.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ro_RO-mihai-medium.tar.bz2

You can use the following code to play with vits-piper-ro_RO-mihai-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ro_RO-mihai-medium/ro_RO-mihai-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ro_RO-mihai-medium/espeak-ng-data",
            tokens="vits-piper-ro_RO-mihai-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Un foc fără lemne se stinge, o lume fără poveste moare.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Un foc fără lemne se stinge, o lume fără poveste moare.

sample audios for different speakers are listed below:

Speaker 0

Russian

This section lists text to speech models for Russian.

vits-piper-ru_RU-denis-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ru/ru_RU/denis/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-denis-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ru_RU-denis-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ru_RU-denis-medium/ru_RU-denis-medium.onnx";
  config.model.vits.tokens = "vits-piper-ru_RU-denis-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ru_RU-denis-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Если курица укусит, ей отрубят голову.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-denis-medium.tar.bz2

You can use the following code to play with vits-piper-ru_RU-denis-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ru_RU-denis-medium/ru_RU-denis-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ru_RU-denis-medium/espeak-ng-data",
            tokens="vits-piper-ru_RU-denis-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Если курица укусит, ей отрубят голову.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Если курица укусит, ей отрубят голову.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ru_RU-dmitri-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ru/ru_RU/dmitri/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-dmitri-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ru_RU-dmitri-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ru_RU-dmitri-medium/ru_RU-dmitri-medium.onnx";
  config.model.vits.tokens = "vits-piper-ru_RU-dmitri-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ru_RU-dmitri-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Если курица укусит, ей отрубят голову.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-dmitri-medium.tar.bz2

You can use the following code to play with vits-piper-ru_RU-dmitri-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ru_RU-dmitri-medium/ru_RU-dmitri-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ru_RU-dmitri-medium/espeak-ng-data",
            tokens="vits-piper-ru_RU-dmitri-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Если курица укусит, ей отрубят голову.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Если курица укусит, ей отрубят голову.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ru_RU-irina-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ru/ru_RU/irina/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-irina-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ru_RU-irina-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ru_RU-irina-medium/ru_RU-irina-medium.onnx";
  config.model.vits.tokens = "vits-piper-ru_RU-irina-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ru_RU-irina-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Если курица укусит, ей отрубят голову.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-irina-medium.tar.bz2

You can use the following code to play with vits-piper-ru_RU-irina-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ru_RU-irina-medium/ru_RU-irina-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ru_RU-irina-medium/espeak-ng-data",
            tokens="vits-piper-ru_RU-irina-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Если курица укусит, ей отрубят голову.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Если курица укусит, ей отрубят голову.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-ru_RU-ruslan-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/ru/ru_RU/ruslan/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-ruslan-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-ru_RU-ruslan-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-ru_RU-ruslan-medium/ru_RU-ruslan-medium.onnx";
  config.model.vits.tokens = "vits-piper-ru_RU-ruslan-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-ru_RU-ruslan-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Если курица укусит, ей отрубят голову.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-ru_RU-ruslan-medium.tar.bz2

You can use the following code to play with vits-piper-ru_RU-ruslan-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-ru_RU-ruslan-medium/ru_RU-ruslan-medium.onnx",
            lexicon="",
            data_dir="vits-piper-ru_RU-ruslan-medium/espeak-ng-data",
            tokens="vits-piper-ru_RU-ruslan-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Если курица укусит, ей отрубят голову.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Если курица укусит, ей отрубят голову.

sample audios for different speakers are listed below:

Speaker 0

Serbian

This section lists text to speech models for Serbian.

vits-piper-sr_RS-serbski_institut-medium

vits-piper-sr_RS-serbski_institut-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/sr/sr_RS/serbski_institut/medium

Number of speakers	Sample rate
2	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sr_RS-serbski_institut-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-sr_RS-serbski_institut-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-sr_RS-serbski_institut-medium/sr_RS-serbski_institut-medium.onnx";
  config.model.vits.tokens = "vits-piper-sr_RS-serbski_institut-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-sr_RS-serbski_institut-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Круг не може постојати без свог центра, а нација не може постојати без својих хероја.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sr_RS-serbski_institut-medium.tar.bz2

You can use the following code to play with vits-piper-sr_RS-serbski_institut-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-sr_RS-serbski_institut-medium/sr_RS-serbski_institut-medium.onnx",
            lexicon="",
            data_dir="vits-piper-sr_RS-serbski_institut-medium/espeak-ng-data",
            tokens="vits-piper-sr_RS-serbski_institut-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Круг не може постојати без свог центра, а нација не може постојати без својих хероја.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Круг не може постојати без свог центра, а нација не може постојати без својих хероја.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Slovak

This section lists text to speech models for Slovak.

vits-piper-sk_SK-lili-medium

vits-piper-sk_SK-lili-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/sk/sk_SK/lili/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sk_SK-lili-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-sk_SK-lili-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-sk_SK-lili-medium/sk_SK-lili-medium.onnx";
  config.model.vits.tokens = "vits-piper-sk_SK-lili-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-sk_SK-lili-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Kto nepozná strach, nepozná vôľu.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sk_SK-lili-medium.tar.bz2

You can use the following code to play with vits-piper-sk_SK-lili-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-sk_SK-lili-medium/sk_SK-lili-medium.onnx",
            lexicon="",
            data_dir="vits-piper-sk_SK-lili-medium/espeak-ng-data",
            tokens="vits-piper-sk_SK-lili-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Kto nepozná strach, nepozná vôľu.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Kto nepozná strach, nepozná vôľu.

sample audios for different speakers are listed below:

Speaker 0

Slovenian

This section lists text to speech models for Slovenian.

vits-piper-sl_SI-artur-medium

vits-piper-sl_SI-artur-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/sl/sl_SI/artur/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sl_SI-artur-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-sl_SI-artur-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-sl_SI-artur-medium/sl_SI-artur-medium.onnx";
  config.model.vits.tokens = "vits-piper-sl_SI-artur-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-sl_SI-artur-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Kto sa nebojí, nie je hlúpy.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sl_SI-artur-medium.tar.bz2

You can use the following code to play with vits-piper-sl_SI-artur-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-sl_SI-artur-medium/sl_SI-artur-medium.onnx",
            lexicon="",
            data_dir="vits-piper-sl_SI-artur-medium/espeak-ng-data",
            tokens="vits-piper-sl_SI-artur-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Kto sa nebojí, nie je hlúpy.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Kto sa nebojí, nie je hlúpy.

sample audios for different speakers are listed below:

Speaker 0

Spanish

This section lists text to speech models for Spanish.

vits-piper-es_ES-carlfm-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/es/es_ES/carlfm/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_ES-carlfm-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-es_ES-carlfm-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-es_ES-carlfm-x_low/es_ES-carlfm-x_low.onnx";
  config.model.vits.tokens = "vits-piper-es_ES-carlfm-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-es_ES-carlfm-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_ES-carlfm-x_low.tar.bz2

You can use the following code to play with vits-piper-es_ES-carlfm-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-es_ES-carlfm-x_low/es_ES-carlfm-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-es_ES-carlfm-x_low/espeak-ng-data",
            tokens="vits-piper-es_ES-carlfm-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-es_ES-davefx-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/es/es_ES/davefx/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_ES-davefx-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-es_ES-davefx-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-es_ES-davefx-medium/es_ES-davefx-medium.onnx";
  config.model.vits.tokens = "vits-piper-es_ES-davefx-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-es_ES-davefx-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_ES-davefx-medium.tar.bz2

You can use the following code to play with vits-piper-es_ES-davefx-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-es_ES-davefx-medium/es_ES-davefx-medium.onnx",
            lexicon="",
            data_dir="vits-piper-es_ES-davefx-medium/espeak-ng-data",
            tokens="vits-piper-es_ES-davefx-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-es_ES-sharvard-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/es/es_ES/sharvard/medium

Number of speakers	Sample rate
2	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_ES-sharvard-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-es_ES-sharvard-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-es_ES-sharvard-medium/es_ES-sharvard-medium.onnx";
  config.model.vits.tokens = "vits-piper-es_ES-sharvard-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-es_ES-sharvard-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_ES-sharvard-medium.tar.bz2

You can use the following code to play with vits-piper-es_ES-sharvard-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-es_ES-sharvard-medium/es_ES-sharvard-medium.onnx",
            lexicon="",
            data_dir="vits-piper-es_ES-sharvard-medium/espeak-ng-data",
            tokens="vits-piper-es_ES-sharvard-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

vits-piper-es_MX-ald-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/es/es_MX/ald/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_MX-ald-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-es_MX-ald-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-es_MX-ald-medium/es_MX-ald-medium.onnx";
  config.model.vits.tokens = "vits-piper-es_MX-ald-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-es_MX-ald-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_MX-ald-medium.tar.bz2

You can use the following code to play with vits-piper-es_MX-ald-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-es_MX-ald-medium/es_MX-ald-medium.onnx",
            lexicon="",
            data_dir="vits-piper-es_MX-ald-medium/espeak-ng-data",
            tokens="vits-piper-es_MX-ald-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-es_MX-claude-high


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/es/es_MX/claude/high

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_MX-claude-high.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-es_MX-claude-high with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-es_MX-claude-high/es_MX-claude-high.onnx";
  config.model.vits.tokens = "vits-piper-es_MX-claude-high/tokens.txt";
  config.model.vits.data_dir = "vits-piper-es_MX-claude-high/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-es_MX-claude-high.tar.bz2

You can use the following code to play with vits-piper-es_MX-claude-high

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-es_MX-claude-high/es_MX-claude-high.onnx",
            lexicon="",
            data_dir="vits-piper-es_MX-claude-high/espeak-ng-data",
            tokens="vits-piper-es_MX-claude-high/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Cuando te encuentres ante una puerta cerrada, no olvides que a veces el destino cierra una puerta para que te desvíes hacia un camino que lleva a una ventana que nunca habrías encontrado por tu cuenta.

sample audios for different speakers are listed below:

Speaker 0

Swahili

This section lists text to speech models for Swahili.

vits-piper-sw_CD-lanfrica-medium

vits-piper-sw_CD-lanfrica-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/sw/sw_CD/lanfrica/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sw_CD-lanfrica-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-sw_CD-lanfrica-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-sw_CD-lanfrica-medium/sw_CD-lanfrica-medium.onnx";
  config.model.vits.tokens = "vits-piper-sw_CD-lanfrica-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-sw_CD-lanfrica-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Mtu mmoja hawezi kuiba mazingira.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sw_CD-lanfrica-medium.tar.bz2

You can use the following code to play with vits-piper-sw_CD-lanfrica-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-sw_CD-lanfrica-medium/sw_CD-lanfrica-medium.onnx",
            lexicon="",
            data_dir="vits-piper-sw_CD-lanfrica-medium/espeak-ng-data",
            tokens="vits-piper-sw_CD-lanfrica-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Mtu mmoja hawezi kuiba mazingira.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Mtu mmoja hawezi kuiba mazingira.

sample audios for different speakers are listed below:

Speaker 0

Swedish

This section lists text to speech models for Swedish.

vits-piper-sv_SE-lisa-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/sv/sv_SE/lisa/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sv_SE-lisa-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-sv_SE-lisa-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-sv_SE-lisa-medium/sv_SE-lisa-medium.onnx";
  config.model.vits.tokens = "vits-piper-sv_SE-lisa-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-sv_SE-lisa-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Liten skog, med många träd";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sv_SE-lisa-medium.tar.bz2

You can use the following code to play with vits-piper-sv_SE-lisa-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-sv_SE-lisa-medium/sv_SE-lisa-medium.onnx",
            lexicon="",
            data_dir="vits-piper-sv_SE-lisa-medium/espeak-ng-data",
            tokens="vits-piper-sv_SE-lisa-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Liten skog, med många träd",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Liten skog, med många träd

sample audios for different speakers are listed below:

Speaker 0

vits-piper-sv_SE-nst-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/sv/sv_SE/nst/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sv_SE-nst-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-sv_SE-nst-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-sv_SE-nst-medium/sv_SE-nst-medium.onnx";
  config.model.vits.tokens = "vits-piper-sv_SE-nst-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-sv_SE-nst-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Liten skog, med många träd";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-sv_SE-nst-medium.tar.bz2

You can use the following code to play with vits-piper-sv_SE-nst-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-sv_SE-nst-medium/sv_SE-nst-medium.onnx",
            lexicon="",
            data_dir="vits-piper-sv_SE-nst-medium/espeak-ng-data",
            tokens="vits-piper-sv_SE-nst-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Liten skog, med många träd",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Liten skog, med många träd

sample audios for different speakers are listed below:

Speaker 0

Turkish

This section lists text to speech models for Turkish.

vits-piper-tr_TR-dfki-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/tr/tr_TR/dfki/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-tr_TR-dfki-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-tr_TR-dfki-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-tr_TR-dfki-medium/tr_TR-dfki-medium.onnx";
  config.model.vits.tokens = "vits-piper-tr_TR-dfki-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-tr_TR-dfki-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-tr_TR-dfki-medium.tar.bz2

You can use the following code to play with vits-piper-tr_TR-dfki-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-tr_TR-dfki-medium/tr_TR-dfki-medium.onnx",
            lexicon="",
            data_dir="vits-piper-tr_TR-dfki-medium/espeak-ng-data",
            tokens="vits-piper-tr_TR-dfki-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz

sample audios for different speakers are listed below:

Speaker 0

vits-piper-tr_TR-fahrettin-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/tr/tr_TR/fahrettin/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-tr_TR-fahrettin-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-tr_TR-fahrettin-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-tr_TR-fahrettin-medium/tr_TR-fahrettin-medium.onnx";
  config.model.vits.tokens = "vits-piper-tr_TR-fahrettin-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-tr_TR-fahrettin-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-tr_TR-fahrettin-medium.tar.bz2

You can use the following code to play with vits-piper-tr_TR-fahrettin-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-tr_TR-fahrettin-medium/tr_TR-fahrettin-medium.onnx",
            lexicon="",
            data_dir="vits-piper-tr_TR-fahrettin-medium/espeak-ng-data",
            tokens="vits-piper-tr_TR-fahrettin-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz

sample audios for different speakers are listed below:

Speaker 0

vits-piper-tr_TR-fettah-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/tr/tr_TR/fettah/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-tr_TR-fettah-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-tr_TR-fettah-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-tr_TR-fettah-medium/tr_TR-fettah-medium.onnx";
  config.model.vits.tokens = "vits-piper-tr_TR-fettah-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-tr_TR-fettah-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-tr_TR-fettah-medium.tar.bz2

You can use the following code to play with vits-piper-tr_TR-fettah-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-tr_TR-fettah-medium/tr_TR-fettah-medium.onnx",
            lexicon="",
            data_dir="vits-piper-tr_TR-fettah-medium/espeak-ng-data",
            tokens="vits-piper-tr_TR-fettah-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Bir evin duvarları, bir adamın sözü, bir kadının gülü kırılmaz

sample audios for different speakers are listed below:

Speaker 0

Ukrainian

This section lists text to speech models for Ukrainian.

vits-piper-uk_UA-lada-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/uk/uk_UA/lada/x_low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-uk_UA-lada-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-uk_UA-lada-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-uk_UA-lada-x_low/uk_UA-lada-x_low.onnx";
  config.model.vits.tokens = "vits-piper-uk_UA-lada-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-uk_UA-lada-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Ви не можете навчити коня, якщо не відвикнете від годівлі.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-uk_UA-lada-x_low.tar.bz2

You can use the following code to play with vits-piper-uk_UA-lada-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-uk_UA-lada-x_low/uk_UA-lada-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-uk_UA-lada-x_low/espeak-ng-data",
            tokens="vits-piper-uk_UA-lada-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Ви не можете навчити коня, якщо не відвикнете від годівлі.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Ви не можете навчити коня, якщо не відвикнете від годівлі.

sample audios for different speakers are listed below:

Speaker 0

vits-piper-uk_UA-ukrainian_tts-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/uk/uk_UA/ukrainian_tts/medium

Number of speakers	Sample rate
3	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-uk_UA-ukrainian_tts-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-uk_UA-ukrainian_tts-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-uk_UA-ukrainian_tts-medium/uk_UA-ukrainian_tts-medium.onnx";
  config.model.vits.tokens = "vits-piper-uk_UA-ukrainian_tts-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-uk_UA-ukrainian_tts-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Ви не можете навчити коня, якщо не відвикнете від годівлі.";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-uk_UA-ukrainian_tts-medium.tar.bz2

You can use the following code to play with vits-piper-uk_UA-ukrainian_tts-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-uk_UA-ukrainian_tts-medium/uk_UA-ukrainian_tts-medium.onnx",
            lexicon="",
            data_dir="vits-piper-uk_UA-ukrainian_tts-medium/espeak-ng-data",
            tokens="vits-piper-uk_UA-ukrainian_tts-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Ви не можете навчити коня, якщо не відвикнете від годівлі.",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Ви не можете навчити коня, якщо не відвикнете від годівлі.

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Vietnamese

This section lists text to speech models for Vietnamese.

vits-piper-vi_VN-25hours_single-low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/vi/vi_VN/25hours_single/low

Number of speakers	Sample rate
1	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-vi_VN-25hours_single-low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-vi_VN-25hours_single-low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-vi_VN-25hours_single-low/vi_VN-25hours_single-low.onnx";
  config.model.vits.tokens = "vits-piper-vi_VN-25hours_single-low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-vi_VN-25hours_single-low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Nước cũ đào gỗ mới, sông cũ chảy nước mới";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-vi_VN-25hours_single-low.tar.bz2

You can use the following code to play with vits-piper-vi_VN-25hours_single-low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-vi_VN-25hours_single-low/vi_VN-25hours_single-low.onnx",
            lexicon="",
            data_dir="vits-piper-vi_VN-25hours_single-low/espeak-ng-data",
            tokens="vits-piper-vi_VN-25hours_single-low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Nước cũ đào gỗ mới, sông cũ chảy nước mới",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Nước cũ đào gỗ mới, sông cũ chảy nước mới

sample audios for different speakers are listed below:

Speaker 0

vits-piper-vi_VN-vais1000-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/vi/vi_VN/vais1000/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-vi_VN-vais1000-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-vi_VN-vais1000-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-vi_VN-vais1000-medium/vi_VN-vais1000-medium.onnx";
  config.model.vits.tokens = "vits-piper-vi_VN-vais1000-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-vi_VN-vais1000-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Nước cũ đào gỗ mới, sông cũ chảy nước mới";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-vi_VN-vais1000-medium.tar.bz2

You can use the following code to play with vits-piper-vi_VN-vais1000-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-vi_VN-vais1000-medium/vi_VN-vais1000-medium.onnx",
            lexicon="",
            data_dir="vits-piper-vi_VN-vais1000-medium/espeak-ng-data",
            tokens="vits-piper-vi_VN-vais1000-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Nước cũ đào gỗ mới, sông cũ chảy nước mới",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Nước cũ đào gỗ mới, sông cũ chảy nước mới

sample audios for different speakers are listed below:

Speaker 0

vits-piper-vi_VN-vivos-x_low


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/vi/vi_VN/vivos/x_low

Number of speakers	Sample rate
65	16000

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-vi_VN-vivos-x_low.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-vi_VN-vivos-x_low with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-vi_VN-vivos-x_low/vi_VN-vivos-x_low.onnx";
  config.model.vits.tokens = "vits-piper-vi_VN-vivos-x_low/tokens.txt";
  config.model.vits.data_dir = "vits-piper-vi_VN-vivos-x_low/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Nước cũ đào gỗ mới, sông cũ chảy nước mới";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-vi_VN-vivos-x_low.tar.bz2

You can use the following code to play with vits-piper-vi_VN-vivos-x_low

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-vi_VN-vivos-x_low/vi_VN-vivos-x_low.onnx",
            lexicon="",
            data_dir="vits-piper-vi_VN-vivos-x_low/espeak-ng-data",
            tokens="vits-piper-vi_VN-vivos-x_low/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Nước cũ đào gỗ mới, sông cũ chảy nước mới",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Nước cũ đào gỗ mới, sông cũ chảy nước mới

sample audios for different speakers are listed below:

Speaker 0

Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

Speaker 6

Speaker 7

Speaker 8

Speaker 9

Speaker 10

Speaker 11

Speaker 12

Speaker 13

Speaker 14

Speaker 15

Speaker 16

Speaker 17

Speaker 18

Speaker 19

Speaker 20

Speaker 21

Speaker 22

Speaker 23

Speaker 24

Speaker 25

Speaker 26

Speaker 27

Speaker 28

Speaker 29

Speaker 30

Speaker 31

Speaker 32

Speaker 33

Speaker 34

Speaker 35

Speaker 36

Speaker 37

Speaker 38

Speaker 39

Speaker 40

Speaker 41

Speaker 42

Speaker 43

Speaker 44

Speaker 45

Speaker 46

Speaker 47

Speaker 48

Speaker 49

Speaker 50

Speaker 51

Speaker 52

Speaker 53

Speaker 54

Speaker 55

Speaker 56

Speaker 57

Speaker 58

Speaker 59

Speaker 60

Speaker 61

Speaker 62

Speaker 63

Speaker 64

Welsh

This section lists text to speech models for Welsh.

vits-piper-cy_GB-bu_tts-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/cy/cy_GB/bu_tts/medium

Number of speakers	Sample rate
7	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cy_GB-bu_tts-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-cy_GB-bu_tts-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-cy_GB-bu_tts-medium/cy_GB-bu_tts-medium.onnx";
  config.model.vits.tokens = "vits-piper-cy_GB-bu_tts-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-cy_GB-bu_tts-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Ni all y gwynt ei hunan ei ddilyn, ac felly mae’n rhaid i’r gŵyr ddod i’r gorwel i weld y llwybr yn gyfarwydd";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cy_GB-bu_tts-medium.tar.bz2

You can use the following code to play with vits-piper-cy_GB-bu_tts-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-cy_GB-bu_tts-medium/cy_GB-bu_tts-medium.onnx",
            lexicon="",
            data_dir="vits-piper-cy_GB-bu_tts-medium/espeak-ng-data",
            tokens="vits-piper-cy_GB-bu_tts-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Ni all y gwynt ei hunan ei ddilyn, ac felly mae’n rhaid i’r gŵyr ddod i’r gorwel i weld y llwybr yn gyfarwydd",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Ni all y gwynt ei hunan ei ddilyn, ac felly mae’n rhaid i’r gŵyr ddod i’r gorwel i weld y llwybr yn gyfarwydd

sample audios for different speakers are listed below:

vits-piper-cy_GB-gwryw_gogleddol-medium


Info about this model	Samples	Andriod APK
C API	Python API

Info about this model

This model is converted from https://huggingface.co/rhasspy/piper-voices/tree/main/cy/cy_GB/gwryw_gogleddol/medium

Number of speakers	Sample rate
1	22050

Model download address

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cy_GB-gwryw_gogleddol-medium.tar.bz2

Android APK

The following table shows the Android TTS Engine APK with this model for sherpa-onnx v1.12.1

ABI	URL	中国镜像
arm64-v8a	Download	下载
armeabi-v7a	Download	下载
x86_64	Download	下载
x86	Download	下载

The source code for the APK can be found at

https://github.com/k2-fsa/sherpa-onnx/tree/master/android/SherpaOnnxTtsEngine

Please refer to the documentation for how to build the APK from source code.

More Android APKs can be found at

https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

C API

You can use the following code to play with vits-piper-cy_GB-gwryw_gogleddol-medium with C API.

#include <stdio.h>
#include <string.h>

#include "sherpa-onnx/c-api/c-api.h"

int main() {
  SherpaOnnxOfflineTtsConfig config;
  memset(&config, 0, sizeof(config));

  config.model.vits.model = "vits-piper-cy_GB-gwryw_gogleddol-medium/cy_GB-gwryw_gogleddol-medium.onnx";
  config.model.vits.tokens = "vits-piper-cy_GB-gwryw_gogleddol-medium/tokens.txt";
  config.model.vits.data_dir = "vits-piper-cy_GB-gwryw_gogleddol-medium/espeak-ng-data";
  config.model.num_threads = 1;
  const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

  int sid = 0; // speaker id
  const char *text = "Ni all y gwynt ei hunan ei ddilyn, ac felly mae’n rhaid i’r gŵyr ddod i’r gorwel i weld y llwybr yn gyfarwydd";

  const SherpaOnnxGeneratedAudio *audio =
      SherpaOnnxOfflineTtsGenerate(tts, text, sid, 1.0);

  SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                      "./test.wav");

  // You need to free the pointers to avoid memory leak in your app
  SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);
  SherpaOnnxDestroyOfflineTts(tts);

  printf("Saved to ./test.wav\n");

  return 0;
}

In the following, we describe how to compile and run the above C example.

Use shared library (dynamic link)

cd /tmp
git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build-shared
cd build-shared

cmake  -DSHERPA_ONNX_ENABLE_C_API=ON  -DCMAKE_BUILD_TYPE=Release  -DBUILD_SHARED_LIBS=ON  -DCMAKE_INSTALL_PREFIX=/tmp/sherpa-onnx/shared  ..

make
make install

You can find required header file and library files inside /tmp/sherpa-onnx/shared.

Assume you have saved the above example file as /tmp/test-piper.c. Then you can compile it with the following command:

gcc   -I /tmp/sherpa-onnx/shared/include   -L /tmp/sherpa-onnx/shared/lib   -lsherpa-onnx-c-api   -lonnxruntime   -o /tmp/test-piper   /tmp/test-piper.c

Now you can run

cd /tmp

# Assume you have downloaded the model and extracted it to /tmp
./test-piper

You probably need to run

# For Linux
export LD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$LD_LIBRARY_PATH

# For macOS
export DYLD_LIBRARY_PATH=/tmp/sherpa-onnx/shared/lib:$DYLD_LIBRARY_PATH

before you run /tmp/test-piper.

Use static library (static link)

Please see the documentation at

https://k2-fsa.github.io/sherpa/onnx/c-api/index.html

Python API

Assume you have installed sherpa-onnx via

pip install sherpa-onnx

and you have downloaded the model from

https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-cy_GB-gwryw_gogleddol-medium.tar.bz2

You can use the following code to play with vits-piper-cy_GB-gwryw_gogleddol-medium

import sherpa_onnx
import soundfile as sf

config = sherpa_onnx.OfflineTtsConfig(
    model=sherpa_onnx.OfflineTtsModelConfig(
        vits=sherpa_onnx.OfflineTtsVitsModelConfig(
            model="vits-piper-cy_GB-gwryw_gogleddol-medium/cy_GB-gwryw_gogleddol-medium.onnx",
            lexicon="",
            data_dir="vits-piper-cy_GB-gwryw_gogleddol-medium/espeak-ng-data",
            tokens="vits-piper-cy_GB-gwryw_gogleddol-medium/tokens.txt",
        ),
        num_threads=1,
    ),
)

if not config.validate():
    raise ValueError("Please check your config")

tts = sherpa_onnx.OfflineTts(config)
audio = tts.generate(text="Ni all y gwynt ei hunan ei ddilyn, ac felly mae’n rhaid i’r gŵyr ddod i’r gorwel i weld y llwybr yn gyfarwydd",
                     sid=0,
                     speed=1.0)

sf.write("test.mp3", audio.samples, samplerate=audio.sample_rate)

Samples

For the following text:

Ni all y gwynt ei hunan ei ddilyn, ac felly mae’n rhaid i’r gŵyr ddod i’r gorwel i weld y llwybr yn gyfarwydd

sample audios for different speakers are listed below:

Keyboard shortcuts

sherpa-onnx text-to-speech samples