mi-gpt  by idootop

Voice assistant for integrating smart speakers with LLMs

created 1 year ago
11,411 stars

Top 4.5% on sourcepulse

GitHubView on GitHub
Project Summary

MiGPT enables users to integrate Xiaomi smart speakers (XiaoAi) with large language models like ChatGPT and Doubao, transforming them into personalized voice assistants. This project targets smart home enthusiasts and users seeking more intelligent and interactive voice control, offering enhanced AI capabilities beyond standard smart speaker functions.

How It Works

MiGPT acts as a bridge, intercepting voice commands from XiaoAi speakers and routing them to LLMs for processing. It leverages a custom agent-based architecture where each smart device can function as an independent agent, capable of sensing and collaborating with others. The system supports long-term memory for conversational context and allows for custom Text-to-Speech (TTS) voices, including those from Doubao, for a more natural interaction.

Quick Start & Requirements

  • Installation: Docker or Node.js.
  • Prerequisites:
    • Xiaomi smart speaker (XiaoAi, Pro recommended).
    • Xiaomi account credentials.
    • For Node.js: Node.js environment.
  • Setup: Configuration of .env and .migpt.js files.
  • Documentation: MiGPT Official Video Tutorial, Usage Documentation

Highlighted Details

  • AI Q&A: Transforms XiaoAi into a knowledgeable assistant.
  • Role-playing: Allows customization of XiaoAi's persona.
  • Streaming responses for near real-time interaction.
  • Long/short-term memory for improved conversational continuity.
  • Customizable TTS voices (e.g., Doubao).
  • Smart home agent capabilities for device automation based on user mood.

Maintenance & Community

The project is actively maintained, with several related community projects offering GUI interfaces and extended functionality (e.g., camera integration). Links to community resources like Discord/Slack are not explicitly provided in the README, but related projects are linked.

Licensing & Compatibility

  • License: MIT License.
  • Compatibility: Suitable for commercial use and linking with closed-source applications due to the permissive MIT license. Does not support other smart speaker brands like Xiaodu or Tmall Genie.

Limitations & Caveats

The project is intended for learning and research and explicitly disclaims responsibility for any misuse, including potential device damage or account bans. Users must comply with local laws. The README notes that the Node.js mode requires manual environment variable setup and passing parameters to MiGPT.create.

Health Check
Last commit

2 months ago

Responsiveness

1 day

Pull Requests (30d)
0
Issues (30d)
0
Star History
478 stars in the last 90 days

Explore Similar Projects

Feedback? Help us improve.