HAPs for VAD + non-streaming speech recognition (HarmonyOS)

This page lists the VAD + non-streaming speech recognition HAPs for sherpa-onnx, one of the deployment frameworks of the Next-gen Kaldi project.
The name of an HAP has the following rule: where
You can download all supported models from https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models

Note about the license The code of Next-gen Kaldi is using Apache-2.0 license. However, we support models from different frameworks. Please check the license of your selected model.

HAP Comment VAD model Non-streaming ASR model
sherpa-onnx-x.y.z-vad_asr-ja-zipformer_reazonspeech.hap It supports only Japanese. It is from https://github.com/reazon-research/ReazonSpeech silero_vad.onnx sherpa-onnx-zipformer-ja-reazonspeech-2024-08-01.tar.bz2
sherpa-onnx-x.y.z-vad_asr-zh_en_ko_ja_yue-sense_voice.hap It supports Chinese, Cantonese, English, Korean, and Japanese (中、英、粤、日、韩5种语音). It is converted from https://github.com/FunAudioLLM/SenseVoice silero_vad.onnx sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
sherpa-onnx-x.y.z-vad_asr-zh-telespeech.hap 支持非常多种中文方言. It is converted from https://github.com/Tele-AI/TeleSpeech-ASR silero_vad.onnx sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04.tar.bz2
sherpa-onnx-x.y.z-vad_asr-th-zipformer.hap It supports only Thai. It is converted from https://huggingface.co/yfyeung/icefall-asr-gigaspeech2-th-zipformer-2024-06-20/tree/main silero_vad.onnx sherpa-onnx-zipformer-thai-2024-06-20.tar.bz2
sherpa-onnx-x.y.z-vad_asr-ko-zipformer.hap It supports only Korean. It is converted from https://huggingface.co/johnBamma/icefall-asr-ksponspeech-zipformer-2024-06-24 silero_vad.onnx sherpa-onnx-zipformer-korean-2024-06-24.tar.bz2
sherpa-onnx-x.y.z-vad_asr-be_de_en_es_fr_hr_it_pl_ru_uk-fast_conformer_ctc_20k.hap It supports 10 languages: Belarusian, German, English, Spanish, French, Croatian, Italian, Polish, Russian, and Ukrainian. It is converted from STT Multilingual FastConformer Hybrid Transducer-CTC Large P&C from NVIDIA/NeMo. Note that only the CTC branch is used. It is trained on ~20000 hours of data. silero_vad.onnx sherpa-onnx-nemo-fast-conformer-transducer-be-de-en-es-fr-hr-it-pl-ru-uk-20k.tar.bz2
sherpa-onnx-x.y.z-vad_asr-en_des_es_fr-fast_conformer_ctc_14288.hap It supports 4 languages: German, English, Spanish, and French . It is converted from STT European FastConformer Hybrid Transducer-CTC Large P&C from NVIDIA/NeMo. Note that only the CTC branch is used. It is trained on 14288 hours of data. silero_vad.onnx sherpa-onnx-nemo-fast-conformer-transducer-en-de-es-fr-14288.tar.bz2
sherpa-onnx-x.y.z-vad_asr-es-fast_conformer_ctc_1424.hap It supports only Spanish. It is converted from STT Es FastConformer Hybrid Transducer-CTC Large P&C from NVIDIA/NeMo. Note that only the CTC branch is used. It is trained on 1424 hours of data. silero_vad.onnx sherpa-onnx-nemo-fast-conformer-transducer-es-1424.tar.bz2
sherpa-onnx-x.y.z-vad_asr-en-fast_conformer_ctc_24500.hap It supports only English. It is converted from STT En FastConformer Hybrid Transducer-CTC Large P&C from NVIDIA/NeMo. Note that only the CTC branch is used. It is trained on 8500 hours of data. silero_vad.onnx sherpa-onnx-nemo-fast-conformer-transducer-en-24500.tar.bz2
sherpa-onnx-x.y.z-vad_asr-zh-zipformer.hap It supports only Chinese. silero_vad.onnx icefall-asr-zipformer-wenetspeech-20230615
sherpa-onnx-x.y.z-vad_asr-zh-paraformer.hap It supports both Chinese and English. silero_vad.onnx sherpa-onnx-paraformer-zh-2023-03-28
sherpa-onnx-x.y.z-vad_asr-en-whisper_tiny.hap It supports only English. silero_vad.onnx sherpa-onnx-whisper-tiny.en
sherpa-onnx-x.y.z-vad_asr-en-moonshine_tiny_int8.hap It supports only English. silero_vad.onnx sherpa-onnx-moonshine-tiny-en-int8
sherpa-onnx-x.y.z-vad_asr-ru-nemo_transducer_giga_am.hap It supports only Russian. silero_vad.onnx sherpa-onnx-nemo-transducer-giga-am-russian-2024-10-24.tar.bz2
Please see also https://github.com/salute-developers/GigaAM
sherpa-onnx-x.y.z-vad_asr-ru-nemo_ctc_giga_am.hap It supports only Russian. silero_vad.onnx sherpa-onnx-nemo-ctc-giga-am-russian-2024-10-24.tar.bz2
Please see also https://github.com/salute-developers/GigaAM
sherpa-onnx-x.y.z-vad_asr-ru-small_zipformer.hap It supports only Russian. silero_vad.onnx sherpa-onnx-small-zipformer-ru-2024-09-18.tar.bz2
sherpa-onnx-x.y.z-vad_asr-ru-zipformer.hap It supports only Russian. silero_vad.onnx sherpa-onnx-zipformer-ru-2024-09-18.tar.bz2


For Chinese users, please visit this address, which replaces huggingface.co with hf-mirror.com

中国用户, 请访问这个地址

sherpa-onnx-1.10.32-vad_asr-zh_en_ko_ja_yue-sense_voice.hap
sherpa-onnx-1.10.32-vad_asr-zh_en-small_paraformer.hap
sherpa-onnx-1.10.32-vad_asr-zh_en-paraformer.hap
sherpa-onnx-1.10.32-vad_asr-zh-zipformer.hap
sherpa-onnx-1.10.32-vad_asr-zh-telespeech.hap
sherpa-onnx-1.10.32-vad_asr-th-zipformer.hap
sherpa-onnx-1.10.32-vad_asr-ru-small_zipformer.hap
sherpa-onnx-1.10.32-vad_asr-ru-zipformer.hap
sherpa-onnx-1.10.32-vad_asr-ru-nemo_transducer_giga_am.hap
sherpa-onnx-1.10.32-vad_asr-ru-nemo_ctc_giga_am.hap
sherpa-onnx-1.10.32-vad_asr-ko-zipformer.hap
sherpa-onnx-1.10.32-vad_asr-ja-zipformer_reazonspeech.hap
sherpa-onnx-1.10.32-vad_asr-es-fast_conformer_ctc_1424.hap
sherpa-onnx-1.10.32-vad_asr-en_de_es_fr-fast_conformer_ctc_14288.hap
sherpa-onnx-1.10.32-vad_asr-en-whisper_tiny.hap
sherpa-onnx-1.10.32-vad_asr-en-moonshine_tiny_int8.hap
sherpa-onnx-1.10.32-vad_asr-en-moonshine_base_int8.hap
sherpa-onnx-1.10.32-vad_asr-en-fast_conformer_ctc_24500.hap
sherpa-onnx-1.10.32-vad_asr-be_de_en_es_fr_hr_it_pl_ru_uk-fast_conformer_ctc_20k.hap