kzr23/VoiceAI — reverse-engineered prompt

Build me a simple full stack VoiceAI app with a frontend and backend that lets someone talk to the app in their browser and get an AI voice experience back. Keep it beginner friendly and easy to run locally, with one clear setup flow that works on Windows, Mac, and Linux using the existing setup scripts.

I want the app to feel polished enough to demo, with a clean page, a clear button to start and stop talking, basic status messages like listening and thinking, and a way to play back or show the response. If voice input is not available, fall back to typed input so the app is still usable. Please wire up the backend so the frontend is not just a mockup, and add sensible environment config, error handling, and a short README with run instructions.

If anything is unclear, make practical choices based on the project name and structure, and look up current docs online if you need to.