HomeProjectsArticlesPK BoardAbout
Explore Projects

LiveKit Agents

Active
GitHub Python Apache-2.0

Description

LiveKit Agents is LiveKit's real-time voice and multimodal agent framework for phone, assistant, and interactive use cases that need low-latency experiences.

Key Features

  • Flexible STT/LLM/TTS/Realtime API integrations — mix and match providers per use case
  • Semantic turn detection using a transformer model to reduce user interruptions
  • Native MCP support — integrate MCP server tools with one line of code
  • Telephony integration via LiveKit SIP stack for phone call agents
  • Built-in job scheduling and dispatch APIs to connect end users to agents
  • Fully open-source stack including LiveKit WebRTC media server

Use Cases

💡 Real-time voice assistants with low-latency speech-to-speech interaction
💡 Multi-agent handoff workflows for customer service and IVR systems
💡 Telephony bots that make and receive phone calls via SIP integration
💡 Multimodal agents that see, hear, and understand in real time
💡 Building and testing voice AI applications with the built-in test framework

Strengths & Limitations

✅ Strengths

  • • Actively maintained, recent updates
  • • High community interest (14.0k stars)
  • • Permissive open-source license (Apache-2.0)
  • • Established track record (2 years in production)

⚠️ Limitations

  • • High issue backlog (775 open issues)

Tags

voice realtime multimodal python

Categories

🤖 Agent Framework

Quick Start

pip install 'livekit-agents[openai,silero,deepgram,cartesia,turn-detector]' && export LIVEKIT_URL=<url> LIVEKIT_API_KEY=<key> LIVEKIT_API_SECRET=<secret> && python my_agent.py start
Visit GitHub Visit Website View Docs

Related Projects

Deepgram Agent API

3.1k · TypeScript
Stale B

Deepgram Agent API is a real-time interface layer for voice agents, combining speech recognition, TTS, and dialog control for phone, assistant, and voice workflow applications.

voicerealtimespeech +1
  • · Real-time speech recognition with streaming audio processing for low-latency voice agent applications
  • · Text-to-speech synthesis with natural voice output for conversational agent responses
  • · Dialog control API for managing multi-turn voice conversations and session state
View on GitHub

Ichigo

2.5k · Python
Stale C

Open-source local realtime voice AI system supporting fully offline real-time voice conversations, suitable for building private voice assistants and voice interaction applications.

pythonvoiceagent +2
  • · Fully offline real-time voice AI system with ASR, LLM, and TTS capabilities
  • · Compact 22M parameter speech tokenizer for multilingual ASR (English + Vietnamese)
  • · Early fusion technique inspired by Meta's Chameleon for speech-language model integration
View on GitHub

Qwen2.5

27.6k · Python
Stale B

Alibaba Qwen Team's open-source LLM series, offering dense and MoE models from 0.6B to 110B parameters with 100+ language support, tool calling, and reasoning — among the strongest open-source models available.

llmqwenalibaba +3
  • · Full-size coverage — Dense 0.6B–72B + MoE 30B-A3B, 235B-A22B for edge to supercomputer deployments
  • · Thinking mode toggle — Seamlessly switch between deep reasoning and efficient chat modes
  • · Million-token context — Native 256K token context, extensible to 1M tokens
View full details → View on GitHub

Related Articles

Voice AgentRealtime APIASR

Voice Agent Architecture: Realtime API and On-Device ASR/TTS

A systematic deep dive into Voice Agent four-layer architecture, latency optimization (streaming ASR, streaming TTS, prompt caching), echo cancellation and interruption handling, on-device ASR/TTS deployment (whisper.cpp, piper, llama.cpp), and a comparison of OpenAI Realtime API, LiveKit Agents, and Pipecat frameworks.

Read more →
语音 AgentLiveKit实时语音

Voice Agent Production Guide: LiveKit Agents from Prototype to Millions of Concurrent Calls

Voice AI agents are the next frontier. LiveKit (11k Stars, powering ChatGPT's Advanced Voice) offers a complete framework. This article breaks down the pipeline and walks through building production-ready voice agents.

Read more →