ruizguille/voice-assistant ? reverse-engineered prompt

Build me a voice assistant app that lets me talk to it out loud and hear it respond back in voice.

I want a simple web app with a clean chat style interface, plus a backend that handles the audio and AI responses. It should listen to my microphone, turn speech into text, send the message to an AI model, then read the reply back to me with text to speech. Please make it work with Groq and Llama 3 for the AI part, and Deepgram for speech recognition and voice output.

It would be great if there’s also a local Python version I can run from the terminal for testing, and a full web version I can open in the browser. Use environment variables for the API keys and make it easy to start locally. If you need to check current docs online for the best setup, go ahead.