search

vlomme / Multi-Tacotron-Voice-Cloning

Phoneme multilingual(Russian-English) voice cloning based on
https://github.com/CorentinJ/Real-Time-Voice-Cloning
Other
379 stars 97 forks source link

Multi-Language Training #17

Open satan17 opened 3 years ago

satan17 commented 3 years ago

Hi,

I'm slightly confused in G2P model. Let's suppose If I need to train a model which specifically translates from Russian to English (Only). Do I still need to add dictionary or train G2P model ?

Also, I'm not able to catch the significance of G2P model here. We have the synthesizer, which is already doing the same work.

Thanks!

vlomme commented 3 years ago

g2p do not translate from Russian to English. It translates graphemes into phonemes. The synthesizer translates any characters into a spectrum.

satan17 commented 3 years ago

Hi,

Sorry, I've mistakenly written the translation. By translation, I intend to ask whether I need to train a g2p model for only English output. I've trained my encoder on other language(German) and I want output in English.

Thanks

vlomme commented 3 years ago

if you only need synthesis in English, then you need to teach it in English. G2p won't help here. The encoder can be trained in any language, it doesn't matter. Use the original implementation