About the OpenAI-compatible API.

[IIIIIllllIIIIIlllll]

I’ve found that many other applications (such as SillyTavern ?) can’t integrate directly with audio.cpp very well, and the compatible API is a bit limited in terms of supported parameters (though I don’t know much about it). How do you all usually use it? Any experience to share? Personally, I just vibe-coded a gateway to solve it, but I think this approach is a bit dumb. Thank you!

[IIIIIllllIIIIIlllll]

The example I ran into today is how to use Qwen TTS deployed with audio.cpp together with SillyTavern. I looked into the code on both sides and found that an ordinary person (i.e., someone who wouldn’t use AI to vibe-code) simply can’t understand how to configure the two to work together. That’s why I’m confused about this compatible API. If I’ve missed anything, please let me know—I’d be very grateful!

[0xShug0]

Maybe this discussion is relevent #289

[0xShug0]

Also #117. The user managed to connect SillyTavern to the audio.cpp server, but some issues in SillyTavern are preventing them from using it properly.

[IIIIIllllIIIIIlllll]

Thanks!