yihong0618/xiaogpt

Play ChatGPT and other LLM with Xiaomi AI Speaker

What it solves

It allows users to integrate Large Language Models (LLMs) like ChatGPT, Gemini, and Llama3 with Xiaomi AI Speakers, enabling the speaker to respond to queries using advanced AI instead of its default system.

How it works

The project acts as a bridge between the Xiaomi speaker and various AI APIs. It monitors the speaker for specific trigger words (e.g., "帮我"), sends the captured query to a selected LLM API, and then uses Text-to-Speech (TTS) to play the AI's response back through the speaker. It supports multiple TTS engines (including Edge, OpenAI, and Azure) and can be configured via CLI arguments or a YAML file.

Who it’s for

Xiaomi AI Speaker owners who want to upgrade their device's intelligence by connecting it to modern LLMs.

Highlights

  • Broad Model Support: Compatible with ChatGPT, Gemini, ChatGLM, Doubao, Moonshot, Llama3, and Qwen.
  • Flexible TTS Options: Supports various high-quality TTS providers like Edge, OpenAI, Azure, and Fish Audio.
  • Advanced Interaction: Features a "continuous conversation" mode and the ability to mute the speaker's default responses.
  • Extensibility: Can integrate with LangChain to enable web search and mathematical operations.

Related

  • Project
  • Project
  • Project
  • Project