I have an audio file containing too much silent segment, and i want to filter the silent segments with torchaudio.functional.vad(). But the funciton can only trim the front silent part of audio, it still remains silent segment in the mid and back. Can someone tell me why? Thanks!
Besides, I want to know the meaning of each parameter in vad(). Beacuse I try to split the audio into many segments with shorter duration, like 0.03s, process each with vad, and concat the process segment together finally, but i don’t know how to set the parameters.
Thanks a lot !