Any good projects for fast voice mode?
I have tried open web UI and it's great when it works but for some reason it's not super reliable or always fast enough for me.
I'm just wondering if there's any simpler tools or projects to integrate voice mode with Ollama models that people have found it and like. Thanks!
7 replies
I don't have an answer, but I'm about to do a similar project on it and would be curious I'd you find something.
It seems like ChatterboxTTS may be a good option, but I have a 5060ti and the default dependencies are being a pain.
I was trying to follow this tutorial, but it is from 2025 and I'm running into other issues unfortunately at the moment.
https://medium.com/@gideont/how-i-got-chatterbox-tts-running-on-an-rtx-5070-pytorch-2-9-cuda-12-8-afc92bb5c10b
That seems to heavy resources for the hardware I have. I did run across these today. Kokoro-82M behind an OpenAI-compatible server
https://localaimaster.com/blog/best-local-tts-models
Kyutai Unmute
https://github.com/kyutai-labs/unmute
I haven't has a chance to start playing with them though
Update, kokoro-fastAPI is awesome: https://github.com/remsky/Kokoro-FastAPI
Only messed with the web interface so far in CPU mode, but it is super fast and high quality
Oh cool, yeah I'm actually pivoting to kokoro now
Good to know. I had to order some hardware but I'll give that my first priority as well.
I actually could not get kokoro to run on my GPU, but to be honest its fast enough for now just running CPU mode that I dont feel like debugging it