ElevenLabs MCP Server

An MCP (Model Context Protocol) server that provides text-to-speech capabilities using ElevenLabs API.

Features

• say: Convert text to speech and save as MP3 file
• read: Read a text file and convert it to speech
• play: Play audio files using the beep library
• set_voice: Change the voice used for generation (memory only)
• get_voices: List available voices and show currently selected one
• history: List available audio files with text summaries

Setup

1. Get your ElevenLabs API key from ElevenLabs
2. Set the environment variable:

   export XI_API_KEY=your_api_key_here
   

Build

go build -o elevenlabs-mcp

Usage

The server runs via stdio and communicates using the MCP protocol:

export XI_API_KEY=your_api_key_here
./elevenlabs-mcp

Tools

say

Convert text to speech and save as MP3.

• text (string, required): Text to convert to speech

read

Read a text file and convert it to speech.

• file_path (string, required): Path to the text file

play

Play an audio file.

• file_path (string, required): Path to the audio file

set_voice

Change the voice used for generation.

• voice_id (string, required): ID of the voice to use

get_voices

List all available voices and show the currently selected one.

• No parameters required

history

List available audio files with text summaries.

• No parameters required

Audio Files

Generated audio files are saved to .xi/<timestamp>-<hex5>.mp3 with corresponding .txt files containing the original text. Files can be replayed using the play tool.

Dependencies

• ElevenLabs API for text-to-speech generation
• Beep library for audio playback
• Mark3labs MCP-Go for MCP server functionality