TTS
TTS copied to clipboard
[Feature request] Tutorial should mention when to use Espeak (or Espeak-ng) and when to use Gruut phonemizer
🚀 Feature Description
It seems that the tutorial deals with Gruut ("If you use phonemes for training and your language is supported here, you don’t need to set your character list.") but some models provided with YourTTS contain Espeak as phonemizer (like the English one). By reading some comments here and there (discussion) some users even report using espeak-ng.
Solution
Tutorial should explain when to use one phonemizer over the other and what the differences in the config.json should be.
Alternative Solutions
Additional context
Hello, if you feel something is missing or would be helpful for the community, please contribute by issuing a PR. Thanks a lot!
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.
@p0p4k now that things are clearer, I will try and write something. According to @WeberJulian in gitter :
No it was for a time since phonemizer wasn't in a compatible licence But we wrote our own espeak wrapper Now it's supported and espeaker is prefered over gruut
So espeak should be used when training from scratch. When fine tuning a model (restore_path) or continuing a previous one (continue_path), then the same phonemizer than used in the original config.json has to be used.
Yeah, you can also finetune from gruut to espeak. By default you should use espeak.
Good to know thank you @WeberJulian !
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.
How to use espeak-ng instead of espeak? I got this error:
Traceback (most recent call last):
File "/home/ubuntu/TTS/vctk_vits_train.py", line 100, in