AI Voice

Give AI a voice that works immediately. One command, zero configuration, speaks in multiple languages using your system’s built-in voice engine.

Quick Start

npx @sylweriusz/mcp-ai-voice

Add to Claude Desktop (~/.claude/mcp_config.json):

{
  "mcpServers": {
    "ai-voice": {
      "command": "npx",
      "args": ["@sylweriusz/mcp-ai-voice"]
    }
  }
}

That’s it. AI can now speak.

What It Does

• Speaks text immediately - AI calls say(), text gets spoken, AI continues working
• Automatic language support - Discovers available voices on startup, picks the best one per language
• Zero setup - Uses your system’s built-in voice (macOS say, Windows SAPI, Linux espeak)
• Works everywhere - macOS, Windows, Linux. No external dependencies.

Usage Examples

// Basic usage
say("Processing complete")

// Language-specific (automatically picks best voice)
say("Hello world", language="en")
say("Witaj świecie", language="pl") 
say("Hola mundo", language="es")

How It Works

The system automatically discovers what languages your computer can speak when it starts up. When AI wants to say something, it picks the best available voice for that language. On macOS this means Enhanced/Premium voices, on Windows it uses SAPI optimization, on Linux it uses language-specific espeak voices.

Platform Support

Requirements

• Node.js 18+
• Any supported OS (macOS/Windows/Linux)
• System voice engine (comes with the OS)

Development

npm install
npm run build
npm start

License

MIT - Use it for whatever you want.

Philosophy: AI needs to express itself vocally sometimes. This makes that happen without hassle.