search

r9y9 / deepvoice3_pytorch

PyTorch implementation of convolutional neural networks-based text-to-speech synthesis models
https://r9y9.github.io/deepvoice3_pytorch/
Other
1.97k stars 485 forks source link

Is it possible to adapt to a new speaker? #163

Closed sdx0112 closed 5 years ago

sdx0112 commented 5 years ago

If I have some audio files from a new speaker, how can I adapt to this speaker using pre-trained model?

r9y9 commented 5 years ago

Please see https://github.com/r9y9/deepvoice3_pytorch#speaker-adaptation.

sdx0112 commented 5 years ago

If I am not taking it wrongly, your instruction is to adapt to VCTK speakers only. My question is to adapt to any arbitrary new speaker.

r9y9 commented 5 years ago

Not limited to VCTK. You'd need to write some code depending on your dataset though. See ljspeech.py for example.