MiGPT-Next enables users to customize responses for Xiaomi smart speakers, allowing for personalized interactions beyond default assistant capabilities. It targets individuals seeking to tailor their smart home experience, offering a flexible platform for custom AI-driven audio output.

How It Works

The project leverages API requests to bridge Xiaomi's smart speaker ecosystem with large language models like OpenAI's GPT. It intercepts user commands, processes them through a configured AI model, and then uses the speaker's Text-to-Speech (TTS) capabilities to deliver the AI-generated response. This approach allows for dynamic, custom replies but can introduce latency and potential conflicts with the speaker's native responses.

Quick Start & Requirements

• Docker: Clone the repository, modify apps/example/config.js with your Xiaomi account details and OpenAI API key, then run docker run -it --rm -v $(pwd)/config.js:/app/config.js idootop/migpt-next:latest.
• Node.js: Install @mi-gpt/next via pnpm install @mi-gpt/next and run the provided TypeScript example, configuring config.js similarly.
• Prerequisites: Xiaomi account credentials, OpenAI API key. Potential need for device flashing to fully resolve "Xiaomi always interrupts" issues.

Highlighted Details

• Supports custom message replies via onMessage function.
• Configuration managed through config.js for speaker and AI settings.
• Removes support for continuous/streaming responses due to latency and interruption issues.
• Provides a workaround for TTS playback issues on specific models by modifying onMessage.

Maintenance & Community

The project is maintained by Del Wang. Links to issue discussions for troubleshooting login failures are provided.

Licensing & Compatibility

Licensed under the MIT License. It is an open-source, non-profit project intended for academic research or personal testing. Commercial use, network attacks, data theft, or system destruction are strictly prohibited. Users agree to terms by downloading or running the project.

Limitations & Caveats

Response latency can be significant, and the system may struggle to interrupt the speaker's native responses. TTS playback can be intermittent, and device status retrieval failures may cause interruptions. A workaround for TTS issues is provided, but some hardware-specific limitations may persist.