search

lifeiteng / vall-e

PyTorch implementation of VALL-E(Zero-Shot Text-To-Speech), Reproduced Demo https://lifeiteng.github.io/valle/index.html
https://lifeiteng.github.io/valle/index.html
Apache License 2.0
2.03k stars 319 forks source link

"Segmentation fault" during data preprocessing #169

Closed KunZhou9646 closed 1 year ago

KunZhou9646 commented 1 year ago

Thanks a lot for your implementation!

During the data pre-processing, I encountered "segmentation fault" issues and still now, don't have a clear idea how to solve it. Could you give me some hints how to solve this issue? Screenshot 2023-10-16 at 12 01 49 PM

lifeiteng commented 1 year ago

Try the two ways:

  1. put blow code into tokenizer.py to get error trace 
    import faulthandler
faulthandler.enable()
  2. try to use CPU when run tokenizer.py to get more detail error info

If you find the error, make a pull request to fix it.

KunZhou9646 commented 1 year ago

It is an incompatible error that indicates espeak was not installed properly. Solved by installing espeak-ng from the source.

Dinxin commented 11 months ago

@KunZhou9646 I ran into the same problem. I also installed espeak-ng from the source. But the problem cannot be solved. So, how to install espeak-ng from the source properly?

KunZhou9646 commented 10 months ago

@Dinxin Can you check if you config espeak-ng properly?