Voice-controlled Reachy Mini robot using OpenAI's Realtime API and Agents SDK. The Realtime API streams voice input/output and enables the model to call Python functions (tools) to control the robot. We use the reachy-mini SDK for robot control.
Supports both OpenAI (gpt-4o-realtime or gpt-4o-mini-realtime) and Azure OpenAI.
# Terminal 1: Start simulator
reachy-mini-daemon --sim
# Terminal 2: Set up and run agent
cp .env.example .env
# Edit .env with your OpenAI API key
uv sync
uv run python -m app.agentapp/
__init__.py
tools.py # Robot motion functions
agent.py # Realtime agent setup
main.py # Original example (not used)
pyproject.toml # Dependencies
.env.example # API key template