voice-agent

Overview ¶

Command voice-agent demonstrates a full voice AI agent using LiveKit.

The agent joins a LiveKit room and responds to human participants using configurable STT, LLM, and TTS providers via omnivoice.

Usage:

export LIVEKIT_URL="wss://your-project.livekit.cloud"
export LIVEKIT_API_KEY="your-api-key"
export LIVEKIT_API_SECRET="your-api-secret"

# STT provider (deepgram, openai, elevenlabs)
export STT_PROVIDER="deepgram"
export DEEPGRAM_API_KEY="your-key"  # or OPENAI_API_KEY, ELEVENLABS_API_KEY

# TTS provider (openai, elevenlabs, deepgram)
export TTS_PROVIDER="openai"
export OPENAI_API_KEY="your-key"    # or ELEVENLABS_API_KEY, DEEPGRAM_API_KEY

# LLM provider
export ANTHROPIC_API_KEY="your-key"

# Optional: Enable avatar (static image in video tile)
export AGENT_AVATAR="true"            # Use default OmniAgent icon
export AGENT_AVATAR="/path/to/avatar.h264"  # Use custom pre-encoded avatar

go run -tags opus ./cmd/voice-agent

Then open the printed URL in your browser to join and talk to the agent.