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
1.46k stars 160 forks source link

Best free cloning app in french ? XTTS V2 ? #217

Closed gandolfi974 closed 5 months ago

gandolfi974 commented 7 months ago

Hello, i'm french. i'm looking for a free cloning app ton integrate in your webui with french voice.

rsxdalv commented 7 months ago

Hi, Bark has some benefits, but it's likely that you are looking for tortoise cloning (XTTS is a variation of tortoise).

As for XTTS, coqui has decided to use a "strong" license that makes it unusable for this project. Unless that has or will change, it will never be in this project.

On Sun, Nov 19, 2023, 12:42 AM gandolfi974 @.***> wrote:

Hello, i'm french. i'm looking for a free cloning app ton integrate in your webui with french voice.

  • Do you have an idea ?
  • is it possible to integrate XTTS coqui V2 ? thanks

— Reply to this email directly, view it on GitHub https://github.com/rsxdalv/tts-generation-webui/issues/217, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTRXI4HQ6RJUCSOD6B2ICLYFDQPDAVCNFSM6AAAAAA7RCUBSGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGAYDANBZGI4DSNY . You are receiving this because you are subscribed to this thread.Message ID: @.***>

gandolfi974 commented 7 months ago

thanks.

rsxdalv commented 6 months ago

thanks.

Sorry for the late follow up, I'm still returning from vacation. Tortoise is not the only one, XTTS is "free" but you are probably going to infringe on the non-commercial part (and that's part of the reason I will not include it in this project). There's also bark and bark voice cloning. Bark is "good" in the sense that it makes a life-like voice. Bark is bad in that it often disregards the prompt. For tortoise and other models (even fairseq), you can get an extra improvement by "upscaling" the output with an RVC model. Arguably that is the best reasonable TTS right now. It ought to be better than what I saw in the youtube video you linked.

gandolfi974 commented 6 months ago

thanks.

rsxdalv commented 6 months ago

thanks.

  • i'm french and XTTS is in french not bark.
  • do you have a link or tutorial for "upscaling" output with RVC model ?

bark purportedly supports French, but I understand that it's performance might not be good enough.

You can see tortoise + RVC search on youtube, there are multiple videos with different examples and depths. https://www.youtube.com/results?search_query=tortoise+rvc

gandolfi974 commented 6 months ago
rsxdalv commented 6 months ago
  • what do you think about this app ? https://github.com/Mangio621/Mangio-RVC-Fork
  • Do you know a free app to synchronize cloning voice with face ? For the moment i use facefusion and i want to make the modified character speak with cloning voice

As for Mangio's fork - I think there were some good changes in it, but I do not follow what's been going on with it. They decided to avoid improving the main project due to language barrier; then the original RVC project got improvements, but Mangio was already too different and could not just "get" these improvements. Then they started asking other devs to help them. I can say that I've seen several RVC forks (I own a small fork to make it installable), and several groups that are working on their RVC. Myself, so far I am happy with the original RVC for this project, but you can install all of the RVCs, they certainly have their pros & cons.

The face sync, I have not done that, I'd recommend searching on youtube, and if you want to discuss it, several discord servers are quite advanced on this topic. Just as an example, camenduuru's server. There must be more but I don't know them. Some stable-diffusion themed AI servers have these discussions as well.

rsxdalv commented 5 months ago

I think I'll close this issue. If there's more questions or discussion, please feel free to reopen.