About masking when training TTS enhanced model

Open WoBuChiTang opened this issue 1 year ago • 2 comments

Hello, thanks for this excellent work!

I am training a TTS model and would like to ask if it is necessary to add any terminator to Y in the prepare_input_target function? Like eog or eos? Can you share the code of TTS model training?

Your answer will help me a lot! Thank you so much!

May 23 '24 07:05 WoBuChiTang

TTS training is just removing the causal masking part of the code. I'm a bit busy right now so not sure when the code will be pushed

May 24 '24 14:05 jasonppy

TTS training is just removing the causal masking part of the code. I'm a bit busy right now so not sure when the code will be pushed

I'm really looking forward to TTS training code

Jul 11 '24 08:07 xiangkanghuang