Edresson/YourTTS

Training yourtts with 48k sampling rate audio

gashishoo7 opened this issue · 2 comments

Hi, I want to train yourTTS model using 48khz sampling rate audio. It would be great help if you could guide me what are the changes are needed. Do I have to change model layers?

It does not require any model architecture changes. You can use the YourTTS recipe by changing the sampling rate.

Thanks a lot for your suggestion. It worked. Sorry for late reply as my training was going on. But I am facing issues with voice cloning and getting some noise in the output audio. In my dataset I have 30 speakers data. How to get good quality in voice cloning? Can you suggest something on that?