search

HarryVolek / PyTorch_Speaker_Verification

PyTorch implementation of "Generalized End-to-End Loss for Speaker Verification" by Wan, Li et al.
BSD 3-Clause "New" or "Revised" License
575 stars 166 forks source link

Why do you use the different vad method for preparing data for d-vector and for uis-rnn? #13

Closed wuqiangch closed 5 years ago

wuqiangch commented 5 years ago

In your code, librosa.effects.split(utter, top_db=30) for d-vector and webrtcvad for uis-rnn.

HarryVolek commented 5 years ago

I use librosa for preprocessing because a VAD isn't needed. You can also try a VAD.