KoljaB
commented
4 months ago Thanks a lot. I share your enhancement recommendations. I looked a lot into appearance. Implementation is hard and requires a lot of changes, but ofc I see the need for this. There already support for local models,no Ollama support yet, but coming soon. Also working on memory which allows remembering user preferences. We also already have a discord server where we discuss exactly these topics regularly: https://discord.gg/f556hqRjpv
This isn't a typical technical issue, and if it offends the issue area of the project, I'll close it. (Sorry for that). Bro, thank you for your great project. I'm not a tech person, and my English isn't that great.
But I particularly like the idea of creating something akin to a truly usable Cortana (Halo) or J.A.R.V.I.S. in the future.
I know you've always been especially skilled in the TTS and ASR aspects. I have an idea. I think most people don't really care about how quickly it can take over (AIoT). In the eyes of ordinary users, it might have three more pressing concerns:
A visually appealing appearance: Live2D, Metahuman? A captivating personality: Local model support like Ollama (after all, everyone knows OpenAI's limitations). Remembering my preferences.
I'm not a very professional PM; mainly, I want to express my gratitude for this project. I think in the future, you could create a Discord community for us to discuss this project. After all, this project already supports good real-time AI voice conversations; it has a solid foundation. To reduce some UI work, it might be worth considering drawing inspiration from “Anime.gf” or simplifying the code from “SillyTavern”. Of course, starting with just a text/voice conversation interface, much like Cortana (Halo), would also be a good choice.