Download
py-xiaozhi is a free, open-source, Python-based AI assistant framework designed for voice interaction, multimodal processing, and connected device control. It supports real-time voice conversations through asynchronous audio streaming and Opus encoding. Users can interact with AI models through spoken commands, camera input, and visual recognition. Its Model Context Protocol integration provides tools for music playback, screenshots, application management, and device operations. The framework supports offline wake-word detection using Sherpa-ONNX. It also communicates through WebSocket and MQTT protocols with encrypted connections and automatic reconnection. Available for Windows, macOS, Linux, and embedded devices, py-xiaozhi supports robotics, IoT applications, and customizable AI assistants.
Features
- Real-time AI voice conversations with Opus audio streaming
- Camera-based visual understanding and image analysis
- MCP tools for applications, music, screenshots, and device control
- Offline wake-word detection with Sherpa-ONNX
- WebSocket and MQTT communication with automatic reconnection
- Cross-platform desktop, Raspberry Pi, and IoT deployment
Project Samples
