How do I change the A.I. model used in the F5-TTS "voice chat" mode?

kaboombomberz commented 1 week ago

Checks

[X] This template is only for usage issues encountered.
[X] I have thoroughly reviewed the project documentation but couldn't find information to solve my problem.
[X] I have searched for existing issues, including closed ones, and couldn't find a solution.
[X] I confirm that I am using English to submit this report in order to facilitate communication.

Environment Details

Windows 10 Pro Python 3.10

Steps to Reproduce

Start F5-TTS Local interface go to "voice chat" tab

✔️ Expected Behavior

Model complies with all the request, including NSFW ones

❌ Actual Behavior

Model denies to engage with NSFW related conversations even after using the 'advanced settings' system prompts.

If there was a way to connect it to run your own model (such as mixtral) it would be great

SWivid commented 1 week ago

Hi @kaboombomberz , that might not be supported with official code as it's sync with hf and other platforms. You could modify the part in https://github.com/SWivid/F5-TTS/blob/c33a83c0094ca4c62c1d193cf457a5e2a2e9d87a/src/f5_tts/infer/infer_gradio.py#L487 with any llm fit in cuz it's currently a cascaded chat system (asr+llm+tts).

kaboombomberz commented 1 week ago

Oh thank you so much! I will try to implement an uncensored model with the code you've provided and let you know if it worked out. I'm guessing I would just have to link in the python run script of the new LLM after the "as app_chat:" correct?

dpp-user commented 3 days ago

i would like to know as well if someone managed to do that, i tried so much time without a success.

SWivid / F5-TTS