SenseVoice 手机独立测试

ASR runtime: k2-fsa/sherpa-onnx, Apache-2.0.
Official source: https://github.com/k2-fsa/sherpa-onnx
Single-thread Emscripten runtime build distributed by:
https://github.com/baijunjie/realtime-translator/tree/main/apps/web/public/sherpa
Upstream build instructions: https://github.com/baijunjie/realtime-translator/blob/main/apps/web/public/sherpa/README.md
The sherpa-onnx license is included as vendor/LICENSE-sherpa-onnx.txt.

ONNX Runtime: Microsoft, MIT License.
https://github.com/microsoft/onnxruntime
SenseVoice model: SenseVoiceSmall by FunAudioLLM / Alibaba.
https://github.com/QwenAudio/SenseVoice
https://huggingface.co/FunAudioLLM/SenseVoiceSmall
Quantized model distribution: k2-fsa/sherpa-onnx, 2024-07-17 int8.
Model terms: https://github.com/modelscope/FunASR/blob/main/MODEL_LICENSE
Model license applies independently from the code license.

Voice activity detector: Silero VAD, MIT License.
https://github.com/snakers4/silero-vad

Test audio is processed on the current device, not uploaded by this page.
Model weights, scripts and runtime binaries are downloaded from this website.