search

Janghyun1230 / Speaker_Verification

Tensorflow implementation of "Generalized End-to-End Loss for Speaker Verification"
MIT License
349 stars 104 forks source link

librosa.effects.trim and librosa.effects.split using default frame_length=2048, hop_length=512, not same to config #23

Open zh794390558 opened 5 years ago

zh794390558 commented 5 years ago

https://github.com/Janghyun1230/Speaker_Verification/blob/170e81ba41f4ec34971a26c90d889a8174d5ea8c/data_preprocess.py#L69