rhasspy / piper

A fast, local neural text to speech system
https://rhasspy.github.io/piper-samples/
MIT License
6.73k stars 494 forks source link

Add support for Hebrew language #538

Open thewh1teagle opened 4 months ago

thewh1teagle commented 4 months ago

What's needed to add support for Hebrew language? The only open tts model in Hebrew currently is mms-tts-heb but the quality is not good.


Some questions related

What's needed to create new good model? How many hours of audio + transcriptions? Do we need them with diacritical marks or no need? Should it be a single speaker? Can we use dataset such as ivrit-ai/audio-labeled - 100+ hours of multiple different speakers, used for create speech to text model so maybe the data not suitable for TTS.

As for the training, how different it should be from training English one? can we fine tune good English model even that hebrew is that different and has diacritical marks?

I really want to add Hebrew support, hope it's the right place to start. Thanks