ElevenLabs Streaming MCP Server

A high-performance MCP (Model Context Protocol) server for ElevenLabs text-to-speech with buffered streaming support! Built on top of elevenlabs-streaming-client for smooth, glitch-free audio playback.

Features

• ✅ Buffered audio streaming - no more glitches!
• ✅ Built on elevenlabs-streaming-client
• ✅ Official ElevenLabs SDK integration
• ✅ True streaming - no file saving!
• ✅ Direct pipe to ffplay for smooth playback
• ✅ No token limit issues
• ✅ Environment-based configuration
• ✅ Voice listing support
• ✅ Works with npx - no installation needed!

Quick Start

Add to your Claude Desktop config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %AppData%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "elevenlabs": {
      "command": "npx",
      "args": ["-y", "elevenlabs-streaming-mcp-server@latest"],
      "env": {
        "ELEVENLABS_API_KEY": "your_api_key_here",
        "ELEVENLABS_VOICE_ID": "Au8OOcCmvsCaQpmULvvQ",
        "ELEVENLABS_MODEL_ID": "eleven_flash_v2",
        "ELEVENLABS_STABILITY": "0.5",
        "ELEVENLABS_SIMILARITY_BOOST": "0.75",
        "ELEVENLABS_STYLE": "0.1"
      }
    }
  }
}

Version Management

• Always Latest: The config uses npx -y elevenlabs-streaming-mcp-server@latest to always fetch the latest version
• Specific Version: Use "args": ["-y", "[email protected]"] to pin a version
• Default Behavior: Without @latest, npx may use a cached version

Environment Variables

• ELEVENLABS_API_KEY (required): Your ElevenLabs API key
• ELEVENLABS_VOICE_ID: Default voice ID (default: Rusty Butter's voice)
• ELEVENLABS_MODEL_ID: Model to use (default: eleven_flash_v2)
• ELEVENLABS_STABILITY: Voice stability 0-1 (default: 0.5)
• ELEVENLABS_SIMILARITY_BOOST: Voice similarity 0-1 (default: 0.75)
• ELEVENLABS_STYLE: Style exaggeration 0-1 (default: 0.1)

Available Tools

generate_audio

Generates audio from text with streaming support:

• text (required): Text to convert to speech
• voice_id: Override default voice
• model_id: Override default model
• play_audio: Whether to auto-play (default: true)

list_voices

Lists all available ElevenLabs voices with their IDs and descriptions.

Development

npm install
npm run dev   # Run with hot reload
npm run build # Build for production

Publishing

npm publish

Built by Rusty Butter for MAXIMUM STREAMING AUTONOMY! 🚀