bigsk1/voice-chat-ai — reverse-engineered prompt

Build me a local voice chat app where I can talk out loud with AI characters through my microphone and hear them answer back in different voices.

I want a simple web UI where I can pick a character, choose the chat provider, choose the voice provider, and switch settings without restarting if possible. It should support local Ollama as well as OpenAI, Anthropic, and xAI for the brain, and support voice options like Spark TTS, OpenAI TTS, ElevenLabs, Kokoro, Typecast, or xAI. If OpenAI realtime is enabled, I want it to feel like a real conversation where I can interrupt the AI and get fast responses.

Please include built in fun characters, plus game and story modes like trivia, escape rooms, and interactive adventures. Let me add new characters with their own personality and voice sample. It should also have a terminal mode, Docker support, a sample env file, and clear setup instructions for Windows, Mac, and Linux. Look up current docs online if you need to.