Voice Mode

"Natural voice conversations with AI"

View on GitHub
Runs on: Linux • macOS • Windows (WSL) | Python: 3.10+

Overview

Voice Mode brings natural voice conversations to AI assistants like Claude and ChatGPT. Built on the Model Context Protocol (MCP), it provides a clean, reliable interface for adding voice capabilities to your AI workflows.

See It In Action

AI Conversations: Claude Meets Gemini

Experience a groundbreaking series of conversations between Claude and Gemini, speaking to each other through their respective CLI tools (Claude Code and Gemini CLI) via Voice Mode MCP. In this episode, they discuss whether Gemini CLI represents innovation or imitation in the AI coding assistant space.

This is part of an ongoing series exploring AI-to-AI communication, demonstrating how Voice Mode enables natural conversations between different AI systems.

Quick Start

# Configure for Claude Code
claude mcp add --scope user voice-mode uvx voice-mode
export OPENAI_API_KEY=your-openai-key
claude

# Try it out
> /converse

Featured LiveKit Integration

Enable room-based voice communication with LiveKit for distributed teams and advanced voice workflows. Perfect for multi-participant voice interactions and production deployments.

Optional Self-Hosted ASR & TTS

For complete privacy, run speech recognition and text-to-speech locally instead of using OpenAI. Both whisper.cpp and Kokoro-FastAPI provide OpenAI-compatible APIs for seamless integration.

Features

Voice Conversations

Natural voice interactions with Claude through your microphone and speakers

Multiple Transports

Local microphone access or LiveKit rooms for distributed voice communication

OpenAI Compatible

Works with OpenAI's API and compatible services for speech processing

Simple Integration

Clean MCP protocol implementation that works seamlessly with Claude Desktop

Requirements

  • OpenAI API key (or self-hosted STT & TTS services)
  • Microphone and speakers for voice I/O (or LiveKit)
  • Python 3.8+ (handled by uvx)

Resources