Voice Agent Prototype

A real-time voice AI agent using LiveKit and Google's Gemini Realtime API for natural conversation.

What It Does

• Real-time bidirectional voice conversation with AI
• Natural speech processing and response generation
• Web-based interface for easy access
• Continuous conversation flow (not just single responses)

Built With

• LiveKit - Real-time audio streaming
• Google Gemini API - AI conversation model
• Flask - Web backend for token generation
• HTML/JavaScript - Browser-based voice interface

Quick Start

Option 1: Docker (Recommended)

Prerequisites: Install Docker Desktop

1. Set up environment:

   cp env.example .env

   Edit .env with your LiveKit and Google Cloud credentials.

2. Run with Docker:

   docker-compose up --build

3. Start conversation:

   • Open http://localhost:5000
   • Click "Join Conversation"
   • Allow microphone access
   • Start talking with the AI agent

Option 2: Local Development

1. Install dependencies:

   pip install -r requirements.txt

2. Set up environment:

   cp env.example .env

   Edit .env with your LiveKit and Google Cloud credentials.

3. Run the application:

   python run_webui.py

4. Start conversation:

   • Open http://localhost:5000
   • Click "Join Conversation"
   • Allow microphone access
   • Start talking with the AI agent

Environment Variables

LIVEKIT_URL=wss://your-livekit-server.livekit.cloud
LIVEKIT_API_KEY=your_livekit_api_key
LIVEKIT_API_SECRET=your_livekit_api_secret
GOOGLE_API_KEY=your_google_api_key