HunterisLive-1/jarvis ? reverse-engineered prompt

Build me a Python voice assistant that listens to my microphone in real time and talks back with natural sounding AI responses using Google Gemini Live.

I want it to be simple to set up, with a requirements file, a .env file for the Google API key, and a single command to run it. It should capture speech, send it to Gemini as I talk, show the transcribed input and output text, and play the assistant’s spoken reply through my speakers with low delay. Keep the code clean and async so it feels responsive.

If you need to check the latest Gemini Live docs or any microphone or audio library details, look them up online and wire it up properly. I’d like the whole thing to feel like a real Jarvis style assistant, just focused on voice in and voice out.