Summary

FireRedChat offers a fully self-hosted framework for real-time, full-duplex voice AI agents. It targets developers and power users seeking customizable, privacy-focused AI agents without external API dependencies, ensuring zero data leakage and complete deployment control.

How It Works

The architecture integrates LiveKit RTC Server for real-time communication, an AI-Agent Bot Server for agent logic, a WebUI for interaction, and Redis for scalability. Dedicated TTS and ASR servers handle speech processing, while a self-hosted LLM (e.g., Ollama, vLLM) powers agent intelligence. This modular, self-hosted approach prioritizes privacy and control.

Quick Start & Requirements

• Installation: Clone repo (git clone --recurse-submodules https://github.com/FireRedTeam/FireRedChat.git).
• Prerequisites: Docker, Docker Compose, self-hosted LLM server (Ollama, vLLM, Dify), optional domain for HTTPS/WSS. Requires specific ports (7880, 3000) and UDP range (50000-60000).
• Setup: Deploy core services (RTC, Redis, WebUI) via docker-compose. Set up ASR, TTS, and LLM services separately. Start AI-Agents Service.
• Links: Ollama guide: https://ollama.com. Other resources like Demo, Paper, Huggingface Overview, Features, Quickstart, and Acknowledgements are mentioned but lack direct URLs in the provided text.

Highlighted Details

• Includes accelerated TTS (text normalization, G2P), ASR (punctuation), personalized pVAD, and compact Turn-Detector (English/Chinese).
• Core components are forks of livekit/agents and livekit/agents-playground.
• Supports multi-node hosting via Redis for scalability.

Maintenance & Community

• Acknowledges contributions from LiveKit and other open-source projects.
• Recent updates (Sept 2025) released key models and services.
• No specific community channels (Discord/Slack) or roadmap links are provided in the README.

Licensing & Compatibility

• License type is not explicitly stated.
• A disclaimer notes content is for "academic purposes only" and "demonstrate technical capabilities," potentially implying non-commercial use restrictions that require clarification.

Limitations & Caveats

• Setup complexity is high due to the need to deploy and manage multiple independent services, including a separate LLM.
• Advanced features like context-aware TTS and audio LLM are listed as "Coming soon."
• The academic disclaimer necessitates verification for commercial deployment.
• Secure connection setup requires manual Nginx configuration.