Mu - Physical Object Control

This project enables natural voice control for physical devices using an ESP32-S3 microcontroller, a Raspberry Pi, ElevenLabs APIs, and a local LLM (tgpt). The ESP32-S3 captures audio commands through a microphone and streams them to the Raspberry Pi, which performs speech-to-text using ElevenLabs. The recognized text is processed by a local LLM to understand user intent, generate a friendly reply, and convert the instruction into a structured physical action (e.g., move left, rotate, open, close). The Raspberry Pi then drives motors or actuators to perform the action and uses ElevenLabs text-to-speech to speak a response back to the user.

The result is a conversational, human-friendly interface for controlling real-world objects, blending embedded hardware, edge AI, and natural language interaction into one seamless system.