search

rsxdalv / tts-generation-webui

TTS Generation Web UI (Bark, MusicGen + AudioGen, Tortoise, RVC, Vocos, Demucs, SeamlessM4T, MAGNet, StyleTTS2, MMS)
https://rsxdalv.github.io/tts-generation-webui/
MIT License
1.71k stars 183 forks source link

RVC training module #124

Open gitihobo opened 1 year ago

gitihobo commented 1 year ago

How can I train rvc inside tts gen webui

rsxdalv commented 1 year ago

There's no GUI yet, although it would be possible with the command line.

gitihobo commented 1 year ago

great news thank you

gitihobo commented 1 year ago

Also Is there any open source technology that lets us upscale audio aside from vocos? I want to recover the voices of older movies

rsxdalv commented 1 year ago

Yes, there are, for example this one: https://kuleshov.github.io/audio-super-res/

But it depends on what do you want to achieve. One goal would be to reconstruct the original with noise etc. Another goal could be to "clean up" the audio. That could be done with denoising + Demucs + RVC.

St33lMouse commented 1 year ago

I have to say your project is really blazing along. I hear good things about RVC, but at the moment we can't train an RVC voice with your system?

I had a look at the rvc stuff online and it is quite confusing. Looks like the original Git is in Japanese? Is that what we would use to train a voice?

rsxdalv commented 1 year ago

The original repo is mostly in Chinese, but has other language readme as well. I would actually recommend searching on YouTube about RVC training.

St33lMouse commented 1 year ago

ok, thanks!

gitihobo commented 1 year ago

what I basically want to do is make it so the music and the voice sounds super high resolution compared to how it was originally, basically more frequencies

gitihobo commented 1 year ago

Is this yet possible?