Voice Agent with OpenAI, LiveKit Agents, and MCP

A voice agent implementation that uses LiveKit Agents Framework to create an intelligent conversational assistant with database capabilities.

Overview

This repository contains the code for a voice agent that leverages:

• LiveKit Agents: For real-time infrastructure and agent framework
• AssemblyAI: For fast and accurate real-time Speech-to-Text
• OpenAI: For language understanding and generation (LLM)
• OpenAI TTS: For giving the agent a voice
• Supabase MCP: For database interactions via Model Context Protocol

The agent follows a Speech-to-Text → LLM → Text-to-Speech pipeline and can interact with Supabase databases through natural language commands.

Running the agent

For to learn how to run the agent, including:

• Prerequisites
• Environment setup
• Implementation details
• Code examples
• Step-by-step instructions

Please refer to the companion blog post: How to build a voice agent with OpenAI, LiveKit Agents, and MCP