search

rhasspy / piper

A fast, local neural text to speech system
https://rhasspy.github.io/piper-samples/
MIT License
5.74k stars 411 forks source link

More natural Chinese voice, Please #278

Open Leroy-X opened 9 months ago

Leroy-X commented 9 months ago

Thanks for such a great project, lightweight and fast.

English is very natural, but Chinese has an English accent and seems unnatural. The segmentation of sentence pauses feels a bit mechanical.

I am very much looking forward to providing a more natural Chinese speech model.

By the way, when I select the Chinese model, mixed reading of Chinese and English is not supported. Thank you so much.

Leroy-X commented 9 months ago

Hi, I know a little bit of python, 6g video memory, can I train the model, I want to try it, but worried. Will out-of-the-box model training tools be available in the future? I think this will allow piper to develop quickly

qt06 commented 9 months ago

These issues in Chinese do have a significant impact on usage.

BornSaint commented 9 months ago

look my answer in #280 you have to finetune it