huangjunsen0406/py-xiaozhi

py-xiaozhi is a Python-based AI assistant runtime that handles voice input/output, MCP server integrations, multimodal task pipelines, and IoT device control from a single codebase. What separates it from the wave of ChatGPT wrappers: the MCP integration layer means you can plug in your own tools without rewriting the core, and the IoT hooks are first-class citizens rather than a demo hack. Cross-platform voice interaction works on Linux, macOS, and Windows without a cloud intermediary for the core loop. The score signal here is unusually high for a GitHub project at this stage, which suggests real traction among builders experimenting with local assistant runtimes. Honest reservation: the documentation is still in Chinese for large portions, which will slow down adoption outside that community. If you read Chinese or are comfortable navigating translated docs, the architecture is worth studying even if you do not ship it directly. -> Best for: AI engineer or solo founder building a self-hosted voice assistant or IoT automation product