LiveKit Self-Hosted Setup

Overview

Open source WebRTC server for real-time audio/video streaming.

Why LiveKit

• Fully open source (Apache 2.0)
• Self-hostable on your own infrastructure
• Python Agents SDK with Ollama integration
• Handles WebRTC complexity
• Supports STT/LLM/TTS pipeline orchestration

Installation

Server

# Download and run
livekit-server --dev

# Or via Docker
docker run -p 7880:7880 livekit/livekit-server --dev

Python SDK

uv add "livekit-agents[openai]~=1.3"
# or
pip install livekit-agents[openai]

Ports Required

• 7880: WebSocket signaling
• 7881: WebRTC over TCP
• 50000-60000/UDP: WebRTC media
• 443: HTTPS/TURN (production)

Ollama Integration

from livekit.plugins import openai

session = AgentSession(
    llm=openai.LLM.with_ollama(
        model="lars-trained",
        base_url="http://localhost:11434/v1"
    )
)

Our Customization

LiveKit is modular - we can: - Use Whisper for STT (keep) - Use LARS/Ollama for LLM (keep) - BYPASS LiveKit TTS - route to our InWorld AI pipeline instead

Links

• GitHub: https://github.com/livekit/livekit
• Agents: https://github.com/livekit/agents
• Docs: https://docs.livekit.io/
• Ollama Guide: https://docs.livekit.io/agents/integrations/llm/ollama/