snakers4 / silero-vad

Silero VAD: pre-trained enterprise-grade Voice Activity Detector
MIT License
4.1k stars 402 forks source link

Support both sox and sox_io backends for in-place audio resampling. #424

Closed yairl closed 7 months ago

yairl commented 7 months ago

Depending on the exact system installed, some systems use the 'sox' backend while others have the 'sox_io' backend.

Documentation from latest torchaudio says this should be sox: https://pytorch.org/audio/stable/generated/torchaudio.list_audio_backends.html?highlight=backends#torchaudio.list_audio_backends

I have seen either sox and sox_io on different machines I use.