Pretrained model is valid only for english speechs?

santi-pdp / segan_pytorch

Speech Enhancement Generative Adversarial Network in PyTorch

MIT License

375 stars 109 forks source link

Pretrained model is valid only for english speechs? #10

Closed qinxunhui closed 5 years ago

qinxunhui commented 5 years ago

if there is another speech but not with English, we must train model again? thanks.

santi-pdp commented 5 years ago

Yes, but loading the English weights might be quite useful as we investigated in this work: https://arxiv.org/abs/1712.06340

theAayushbajaj commented 5 years ago

What sampling rate and other voice params should the audio be (in other language) to get a result on pretrained weights? I tried using pretrained weights and the enhanced audio file is blank.