audio-webui
audio-webui copied to clipboard
[FEATURE REQUEST] Add VITS2 support
Please add support to this implementation of VITS2:
https://github.com/p0p4k/vits2_pytorch/
There are English and Russian weights in that repository already. Even though the voices have no "emotion", the speech sounds natural and it hallucinates way, way less than Bark, Tacotron and similar things I tried in the past. Plus it's blazing fast.
And most importantly, it includes training code.
There is also valle-x now. https://github.com/Plachtaa/VALL-E-X