search

VinAIResearch / XPhoneBERT

XPhoneBERT: A Pre-trained Multilingual Model for Phoneme Representations for Text-to-Speech (INTERSPEECH 2023)
MIT License
297 stars 36 forks source link

Chinese data about tones #18

Closed russell-shu closed 8 months ago

russell-shu commented 8 months ago

hi, have you taken chinese pinyin tones into consider when training the model . for example aishell3, ['chi2','qi3','hong2','ying1','qiang1'], were you process the data only in the way ['chi','qi','hong','ying','qiang']?

thanks in advance.

thelinhbkhn2014 commented 8 months ago

We just used the CharsiuG2P to convert text to phonemes. The authors are Chinese, so I believe they were likely to care about it. Please check it at their repo.