gitmylo / audio-webui

A webui for different audio related Neural Networks
MIT License
1.06k stars 100 forks source link

Where to place model? #6

Closed Ph0rk0z closed 1 year ago

Ph0rk0z commented 1 year ago

I have downloaded the model separately but I'm not sure on the folder structure. I tried putting it in suno--bark but it still tries to download the model.

gitmylo commented 1 year ago

Bark models are saved in ~/.cache/suno/bark_v0 by default. Where ~ is C:\Users\username on windows. I might change it later to use the models directory from the webui. But I didn't yet because bark works slightly differently and I'll have to monkeypatch more bark things.

Ph0rk0z commented 1 year ago

I got it working and wow. When it wants to it clones better than tortoise. Faster too. Just unfortunately there is no consistency and the next sentence can be another voice. I set the wav temp to 1 and it would get back almost indistinguishable copies.

gitmylo commented 1 year ago

Right, bark can switch voices, this is also an issue with the official speaker prompt files

Ph0rk0z commented 1 year ago

I'm having good luck with .8 temperature.. for that new RVC, any recommendations for models, esp RVC model. I see the TTS models are pretty much coqui.

gitmylo commented 1 year ago

I'm having good luck with .8 temperature.. for that new RVC, any recommendations for models, esp RVC model. I see the TTS models are pretty much coqui.

some that i know of https://huggingface.co/QuickWick/Music-AI-Voices/tree/main https://discord.gg/aihub