This project provides a commercial-grade backend service for the "小智 AI" voice-interactive chatbot. It targets developers and businesses seeking a scalable, high-concurrency solution for natural human-computer dialogue, integrating with various LLMs and multi-terminal devices via the MCP protocol. The service offers a complete ecosystem for commercial deployment, including user management, payment integration, and flexible model access, aiming for low cost and ease of use.

How It Works

The backend utilizes the MCP protocol to connect diverse clients (ESP32, Python, Android) and supports multiple AI models for ASR, TTS, LLM, and image analysis through API calls or local deployments. It features distinct dialogue modes (auto, manual, realtime) with interruption capabilities, enabling efficient and natural interactions. The architecture is designed for high concurrency, supporting over 3000 online users per machine and offering distributed scalability.

Quick Start & Requirements

• Primary install: Download Release binaries or deploy via Docker. Configuration involves setting up .config.yaml with WebSocket, OTA, and model service details.
• Prerequisites: Go 1.24.2+, specific CGO/Opus libraries for Windows builds, and API keys for integrated AI services.
• Links: Releases page, MCP protocol README, Swagger documentation.

Highlighted Details

• High concurrency: Scalable to millions of users.
• Commercial features: User system, payment integration, management backend, MQTT, Dify workflow support (commercial).
• Model flexibility: Integrates ASR (Doubao), TTS (EdgeTTS/Doubao), LLM (OpenAI API, Ollama), and image analysis (Zhipu).
• Client compatibility: ESP32, Python, Android via MCP.
• OTA firmware distribution.

Maintenance & Community

The project welcomes community contributions via Issues and PRs. Custom development services are available. Specific details on active contributors, roadmap, or community channels like Discord/Slack are not provided in the README.

Licensing & Compatibility

The project is licensed under the "Xiaozhi-server-go Open Source License," based on Apache 2.0 Enhanced. While the base license is permissive, certain advanced features are designated as "commercial," suggesting potential licensing variations or tiered offerings.

Limitations & Caveats

Several advanced features, including database support beyond SQLite (MySQL, PostgreSQL), MQTT, and Dify workflow integration, are explicitly marked as "commercial," indicating they may require separate licensing or are part of a paid offering. The README does not detail specific performance benchmarks beyond concurrency claims or mention known bugs or deprecations.