FastSpeech2 icon indicating copy to clipboard operation
FastSpeech2 copied to clipboard

to support more symbols for own dataset

Open everschen opened this issue 3 years ago • 1 comments

I use my own lexicon and did MFA train my own dataset, after preprocess.py during training, the issue #13 was found, after add the missing symbols to pinyin.py, the issue has been fixed.

everschen avatar May 22 '22 09:05 everschen

I use my own lexicon and did MFA train my own dataset, after preprocess.py during training, the issue #13 was found, after add the missing symbols to pinyin.py, the issue has been fixed.

Hi am facing the same issue RuntimeError : The size of tensor.. when i used the code that was mentioned to print the phones, i got many list of phones. Now i do not know where to add them in pinyin.py as there are two list initial and final. Please guide me

Lakhjeet1082 avatar May 11 '24 16:05 Lakhjeet1082