LiveKit Agent

Prerequisites

• Python > 3.10
• venv (comes with Python)
• Internet connection for downloading models and dependencies

Setup

python -m venv venv
source venv/bin/activate          # On Windows: venv\Scripts\activate
pip install -r requirements.txt
python agent.py download-files
python agent.py console

Usage & Metrics Collection

1. Start the agent in console mode:

   python agent.py console
   

2. Interact with the agent through voice or text input

3. Stop the agent by pressing Ctrl + C to gracefully shutdown and save metrics

4. View collected metrics in the generated file:

   cat livekit_agent_metrics.json
   

Metrics Data

The agent automatically collects comprehensive metrics including:

• LLM Metrics: Token usage, response times (TTFT), model performance
• TTS Metrics: Character count, audio generation times (TTFB), audio duration
• STT Metrics: Audio processing duration, transcription accuracy
• Session Data: Conversation turns, timestamps, duration
• Usage Summary: Official LiveKit usage statistics

Metrics File Structure

{
  "sessions": [
    {
      "session_id": "uuid",
      "duration": 12.22,
      "total_llm_calls": 3,
      "total_llm_tokens_input": 71,
      "total_llm_tokens_output": 51,
      "total_tts_calls": 1,
      "total_tts_characters": 56,
      "total_stt_calls": 2,
      "total_stt_duration": 9.95,
      "conversation_turns": 7,
      "transcripts": [...],
      "usage_summary": {...}
    }
  ]
}

Key Features

• ✅ Automatic metrics collection using LiveKit's official metrics system
• ✅ Real-time monitoring with console output during sessions
• ✅ Persistent storage in JSON format for analysis
• ✅ Conversation tracking with full transcript history
• ✅ Usage analytics for cost estimation and optimization
• ✅ Error-safe serialization handles all LiveKit metric types

Troubleshooting

If you encounter any issues:

1. Ensure all dependencies are installed: pip install -r requirements.txt
2. Check that your environment variables are set correctly
3. Verify internet connection for model downloads
4. Make sure you have proper API keys configured