Paraformer models

Hint

Please refer to Installation to install sherpa-onnx before you read this section.

csukuangfj/sherpa-onnx-streaming-paraformer-bilingual-zh-en (Chinese + English)

Note

This model does not support timestamps. It is a bilingual model, supporting both Chinese and English. (支持普通话、河南话、天津话、四川话等方言)

This model is converted from

https://www.modelscope.cn/models/damo/speech_paraformer_asr_nat-zh-cn-16k-common-vocab8404-online/summary

The code for converting can be found at

https://huggingface.co/csukuangfj/streaming-paraformer-zh

In the following, we describe how to download it and use it with sherpa-onnx.

Download the model

Please use the following commands to download it.

cd /path/to/sherpa-onnx

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2

# For Chinese users
# wget https://hub.nuaa.cf/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2

tar xvf sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2

Please check that the file sizes of the pre-trained models are correct. See the file sizes of *.onnx files below.

sherpa-onnx-streaming-paraformer-bilingual-zh-en fangjun$ ls -lh *.onnx
-rw-r--r--  1 fangjun  staff    68M Aug 14 09:53 decoder.int8.onnx
-rw-r--r--  1 fangjun  staff   218M Aug 14 09:55 decoder.onnx
-rw-r--r--  1 fangjun  staff   158M Aug 14 09:54 encoder.int8.onnx
-rw-r--r--  1 fangjun  staff   607M Aug 14 09:57 encoder.onnx

Decode a single wave file

Hint

It supports decoding only wave files of a single channel with 16-bit encoded samples, while the sampling rate does not need to be 16 kHz.

fp32

The following code shows how to use fp32 models to decode a wave file:

cd /path/to/sherpa-onnx

./build/bin/sherpa-onnx \
  --tokens=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt \
  --paraformer-encoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.onnx \
  --paraformer-decoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.onnx \
  ./sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav

Note

Please use ./build/bin/Release/sherpa-onnx.exe for Windows.

Caution

If you use Windows and get encoding issues, please run:

CHCP 65001

in your commandline.

You should see the following output:

/Users/fangjun/open-source/sherpa-onnx/sherpa-onnx/csrc/parse-options.cc:Read:361 ./build/bin/sherpa-onnx --tokens=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt --paraformer-encoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.onnx --paraformer-decoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.onnx ./sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav 

OnlineRecognizerConfig(feat_config=FeatureExtractorConfig(sampling_rate=16000, feature_dim=80), model_config=OnlineModelConfig(transducer=OnlineTransducerModelConfig(encoder="", decoder="", joiner=""), paraformer=OnlineParaformerModelConfig(encoder="./sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.onnx", decoder="./sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.onnx"), tokens="./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt", num_threads=1, debug=False, provider="cpu", model_type=""), lm_config=OnlineLMConfig(model="", scale=0.5), endpoint_config=EndpointConfig(rule1=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=2.4, min_utterance_length=0), rule2=EndpointRule(must_contain_nonsilence=True, min_trailing_silence=1.2, min_utterance_length=0), rule3=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=0, min_utterance_length=20)), enable_endpoint=True, max_active_paths=4, context_score=1.5, decoding_method="greedy_search")
./sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav
Elapsed seconds: 2.2, Real time factor (RTF): 0.21
昨天是 monday today day is 零八二 the day after tomorrow 是星期三
{"is_final":false,"segment":0,"start_time":0.0,"text":"昨天是 monday today day is 零八二 the day after tomorrow 是星期三","timestamps":"[]","tokens":["昨","天","是","mon@@","day","today","day","is","零","八","二","the","day","after","tom@@","or@@","row","是","星","期","三"]}

int8

The following code shows how to use int8 models to decode a wave file:

cd /path/to/sherpa-onnx

./build/bin/sherpa-onnx \
  --tokens=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt \
  --paraformer-encoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.int8.onnx \
  --paraformer-decoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.int8.onnx \
  ./sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav

Note

Please use ./build/bin/Release/sherpa-onnx.exe for Windows.

Caution

If you use Windows and get encoding issues, please run:

CHCP 65001

in your commandline.

You should see the following output:

/Users/fangjun/open-source/sherpa-onnx/sherpa-onnx/csrc/parse-options.cc:Read:361 ./build/bin/sherpa-onnx --tokens=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt --paraformer-encoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.int8.onnx --paraformer-decoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.int8.onnx ./sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav 

OnlineRecognizerConfig(feat_config=FeatureExtractorConfig(sampling_rate=16000, feature_dim=80), model_config=OnlineModelConfig(transducer=OnlineTransducerModelConfig(encoder="", decoder="", joiner=""), paraformer=OnlineParaformerModelConfig(encoder="./sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.int8.onnx", decoder="./sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.int8.onnx"), tokens="./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt", num_threads=1, debug=False, provider="cpu", model_type=""), lm_config=OnlineLMConfig(model="", scale=0.5), endpoint_config=EndpointConfig(rule1=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=2.4, min_utterance_length=0), rule2=EndpointRule(must_contain_nonsilence=True, min_trailing_silence=1.2, min_utterance_length=0), rule3=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=0, min_utterance_length=20)), enable_endpoint=True, max_active_paths=4, context_score=1.5, decoding_method="greedy_search")
./sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav
Elapsed seconds: 1.6, Real time factor (RTF): 0.15
昨天是 monday today day is 零八二 the day after tomorrow 是星期三
{"is_final":false,"segment":0,"start_time":0.0,"text":"昨天是 monday today day is 零八二 the day after tomorrow 是星期三","timestamps":"[]","tokens":["昨","天","是","mon@@","day","today","day","is","零","八","二","the","day","after","tom@@","or@@","row","是","星","期","三"]}

Real-time speech recognition from a microphone

cd /path/to/sherpa-onnx

./build/bin/sherpa-onnx-microphone \
  --tokens=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt \
  --paraformer-encoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.int8.onnx \
  --paraformer-decoder=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.int8.onnx

Hint

If your system is Linux (including embedded Linux), you can also use sherpa-onnx-alsa to do real-time speech recognition with your microphone if sherpa-onnx-microphone does not work for you.

csukuangfj/sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en (Chinese + Cantonese + English)

Note

This model does not support timestamps. It is a trilingual model, supporting both Chinese and English. (支持普通话、粤语、河南话、天津话、四川话等方言)

This model is converted from

https://modelscope.cn/models/dengcunqin/speech_paraformer-large_asr_nat-zh-cantonese-en-16k-vocab8501-online/files

You can find the conversion code after downloading and unzipping the model.

In the following, we describe how to download it and use it with sherpa-onnx.

Download the model

Please use the following commands to download it.

cd /path/to/sherpa-onnx

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en.tar.bz2

# For Chinese users
# wget https://hub.nuaa.cf/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en.tar.bz2

tar xvf sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en.tar.bz2

Please check that the file sizes of the pre-trained models are correct. See the file sizes of *.onnx files below.

sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en fangjun$ ls -lh *.onnx
-rw-r--r--  1 fangjun  staff    69M Feb 29 19:44 decoder.int8.onnx
-rw-r--r--  1 fangjun  staff   218M Feb 29 19:44 decoder.onnx
-rw-r--r--  1 fangjun  staff   159M Feb 29 19:44 encoder.int8.onnx
-rw-r--r--  1 fangjun  staff   607M Feb 29 19:44 encoder.onnx

Decode a single wave file

Hint

It supports decoding only wave files of a single channel with 16-bit encoded samples, while the sampling rate does not need to be 16 kHz.

fp32

The following code shows how to use fp32 models to decode a wave file:

cd /path/to/sherpa-onnx

./build/bin/sherpa-onnx \
  --tokens=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/tokens.txt \
  --paraformer-encoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/encoder.onnx \
  --paraformer-decoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/decoder.onnx \
  ./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/test_wavs/1.wav

Note

Please use ./build/bin/Release/sherpa-onnx.exe for Windows.

Caution

If you use Windows and get encoding issues, please run:

CHCP 65001

in your commandline.

You should see the following output:

/Users/fangjun/open-source/sherpa-onnx/sherpa-onnx/csrc/parse-options.cc:Read:361 ./build/bin/sherpa-onnx --tokens=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/tokens.txt --paraformer-encoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/encoder.int8.onnx --paraformer-decoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/decoder.int8.onnx ./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/test_wavs/1.wav 

OnlineRecognizerConfig(feat_config=FeatureExtractorConfig(sampling_rate=16000, feature_dim=80), model_config=OnlineModelConfig(transducer=OnlineTransducerModelConfig(encoder="", decoder="", joiner=""), paraformer=OnlineParaformerModelConfig(encoder="./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/encoder.int8.onnx", decoder="./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/decoder.int8.onnx"), wenet_ctc=OnlineWenetCtcModelConfig(model="", chunk_size=16, num_left_chunks=4), zipformer2_ctc=OnlineZipformer2CtcModelConfig(model=""), tokens="./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/tokens.txt", num_threads=1, debug=False, provider="cpu", model_type=""), lm_config=OnlineLMConfig(model="", scale=0.5), endpoint_config=EndpointConfig(rule1=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=2.4, min_utterance_length=0), rule2=EndpointRule(must_contain_nonsilence=True, min_trailing_silence=1.2, min_utterance_length=0), rule3=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=0, min_utterance_length=20)), enable_endpoint=True, max_active_paths=4, hotwords_score=1.5, hotwords_file="", decoding_method="greedy_search", blank_penalty=0)
./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/test_wavs/1.wav
Elapsed seconds: 0.98, Real time factor (RTF): 0.16
有无人知道湾仔活道系点去
{ "text": "有无人知道湾仔活道系点去", "tokens": [ "有", "无", "人", "知", "道", "湾", "仔", "活", "道", "系", "点", "去" ], "timestamps": [  ], "ys_probs": [  ], "lm_probs": [  ], "context_scores": [  ], "segment": 0, "start_time": 0.00, "is_final": false}

int8

The following code shows how to use int8 models to decode a wave file:

cd /path/to/sherpa-onnx

./build/bin/sherpa-onnx \
  --tokens=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/tokens.txt \
  --paraformer-encoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/encoder.int8.onnx \
  --paraformer-decoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/decoder.int8.onnx \
  ./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/test_wavs/1.wav

Note

Please use ./build/bin/Release/sherpa-onnx.exe for Windows.

Caution

If you use Windows and get encoding issues, please run:

CHCP 65001

in your commandline.

You should see the following output:

/Users/fangjun/open-source/sherpa-onnx/sherpa-onnx/csrc/parse-options.cc:Read:361 ./build/bin/sherpa-onnx --tokens=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/tokens.txt --paraformer-encoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/encoder.int8.onnx --paraformer-decoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/decoder.int8.onnx ./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/test_wavs/1.wav 

OnlineRecognizerConfig(feat_config=FeatureExtractorConfig(sampling_rate=16000, feature_dim=80), model_config=OnlineModelConfig(transducer=OnlineTransducerModelConfig(encoder="", decoder="", joiner=""), paraformer=OnlineParaformerModelConfig(encoder="./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/encoder.int8.onnx", decoder="./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/decoder.int8.onnx"), wenet_ctc=OnlineWenetCtcModelConfig(model="", chunk_size=16, num_left_chunks=4), zipformer2_ctc=OnlineZipformer2CtcModelConfig(model=""), tokens="./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/tokens.txt", num_threads=1, debug=False, provider="cpu", model_type=""), lm_config=OnlineLMConfig(model="", scale=0.5), endpoint_config=EndpointConfig(rule1=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=2.4, min_utterance_length=0), rule2=EndpointRule(must_contain_nonsilence=True, min_trailing_silence=1.2, min_utterance_length=0), rule3=EndpointRule(must_contain_nonsilence=False, min_trailing_silence=0, min_utterance_length=20)), enable_endpoint=True, max_active_paths=4, hotwords_score=1.5, hotwords_file="", decoding_method="greedy_search", blank_penalty=0)
./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/test_wavs/1.wav
Elapsed seconds: 0.84, Real time factor (RTF): 0.14
有无人知道湾仔活道系点去
{ "text": "有无人知道湾仔活道系点去", "tokens": [ "有", "无", "人", "知", "道", "湾", "仔", "活", "道", "系", "点", "去" ], "timestamps": [  ], "ys_probs": [  ], "lm_probs": [  ], "context_scores": [  ], "segment": 0, "start_time": 0.00, "is_final": false}

Real-time speech recognition from a microphone

cd /path/to/sherpa-onnx

./build/bin/sherpa-onnx-microphone \
  --tokens=./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt \
  --paraformer-encoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/encoder.int8.onnx \
  --paraformer-decoder=./sherpa-onnx-streaming-paraformer-trilingual-zh-cantonese-en/decoder.int8.onnx

Hint

If your system is Linux (including embedded Linux), you can also use sherpa-onnx-alsa to do real-time speech recognition with your microphone if sherpa-onnx-microphone does not work for you.