playbooks

home / mcp / elevenlabs mcp server

ElevenLabs MCP Server

Provides an MCP server that connects ElevenLabs TTS and audio APIs to generate speech, clone voices, and transcribe audio.

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "garblesnarff-elevenlabs-mcp": {
      "command": "uvx",
      "args": [
        "elevenlabs-mcp"
      ],
      "env": {
        "ELEVENLABS_API_KEY": "<insert-your-api-key-here>",
        "ELEVENLABS_MCP_BASE_PATH": "/path/to/mcp_base"
      }
    }
  }
}

You can run an MCP server for ElevenLabs that lets clients like Claude Desktop and other MCP-enabled tools generate speech, clone voices, transcribe audio, and perform audio processing tasks through ElevenLabs APIs. This guide covers how to use the server with your MCP client, how to install and start it, and practical tips for configuration and troubleshooting.

How to use

To interact with ElevenLabs via MCP, configure your MCP client with the provided server settings and supply your ElevenLabs API key. The server runs locally through a command that starts the MCP listener, allowing your client to request speech synthesis, voice cloning, transcription, and related capabilities.

How to install

Prerequisites: you need a working Python environment and the MCP client tooling. You also need an ElevenLabs API key with access to the text-to-speech and audio processing features.

# Step 1: Install the MCP runtime tooling
# Ensure you have uv installed according to your platform
# Step 2: Prepare to run the ElevenLabs MCP server via MCP config

Additional features and notes

Optional environment configuration lets you specify a base path for outputs and inputs. If you want to direct where the MCP server should read inputs or write outputs, set ELEVENLABS_MCP_BASE_PATH in your environment for the Claude Desktop configuration.

Troubleshooting tips: if you encounter timeouts during long-running operations like voice design or audio isolation, these may occur in development tooling but should not occur when using a client like Claude Desktop. If you see a spawn error for uvx, verify the absolute path to uvx and adjust the command accordingly.

Windows users: enable Developer Mode in Claude Desktop to use the MCP server. Open Help > Enable Developer Mode from the Claude application menu if required.

Configuration and start command

{
  "mcpServers": {
    "ElevenLabs": {
      "command": "uvx",
      "args": ["elevenlabs-mcp"],
      "env": {
        "ELEVENLABS_API_KEY": "<insert-your-api-key-here>"
      }
    }
  }
}

Troubleshooting

Logs you may check if you run into issues are typically found where the MCP client writes logs. If you experience timeouts or spawn errors, confirm the path to the uvx executable and ensure the environment variables are present and correctly referenced.

Available tools

synthesize

Generate speech from text using ElevenLabs TTS via MCP

clone_voice

Clone or modify a voice profile for output variation

transcribe

Transcribe audio input to text with speaker separation when provided

soundscape

Create audio environments or soundscapes as described in prompts