PyTorch implementation of VALL-E(Zero-Shot Text-To-Speech), Reproduced Demo https://lifeiteng.github.io/valle/index.html
1.99k
stars
320
forks
source link
I trained a Chinese model, and when synthesizing long speech, the effect may deteriorate, even with pronunciation errors. Why is this? #155
Closed
zhiCharon closed 1 year ago
inference shoud be consist with training.