Skip to main content
Guide1 min read

ElevenLabs MCP server: text-to-speech & audio in Claude (2026)

Generate speech, clone voices and transcribe audio from Claude or Cursor with the official ElevenLabs MCP server. API key, uvx/npx setup, and fixes.

The official ElevenLabs MCP server brings text-to-speech, voice cloning, audio transcription and more into an AI client — so Claude can narrate text, generate voiceovers or transcribe a file on request. Here's the setup.

Step 1 — get an API key

In your ElevenLabs account, open the profile/API settings and copy your API key. Generation consumes plan credits.

Step 2 — configure the client

Via uvx (Python):

{
  "mcpServers": {
    "elevenlabs": {
      "command": "uvx",
      "args": ["elevenlabs-mcp"],
      "env": { "ELEVENLABS_API_KEY": "your_api_key" }
    }
  }
}

Requires Python 3.10+ with uv. An npm package and a Docker image are also published. Restart the client; see MCP config file location.

What it gives your AI

Text-to-speech in many voices, voice cloning, speech-to-text transcription, sound-effect generation and audio processing — driven from chat.

Verify

Ask: "Convert this paragraph to speech in a calm narrator voice and save the MP3." A returned audio file confirms it works.

Common problems

  • 401 — wrong API key.
  • Out of credits — generation uses plan credits; check your balance.
  • uvx: command not found — install uv, or use the npm/Docker option.
  • No tools in client — see Cursor MCP not working.

Going further

ElevenLabs pairs with a filesystem server (save audio) and YouTube transcript / Higgsfield for full media workflows. Browse the content category or curated loadouts. Only clone voices you have rights to.

Loadout

Build your AI agent loadout

The directory of MCP servers and AI agents that actually work. Pick the right loadout for Slack, Postgres, GitHub, Figma and 20+ integrations — with install commands ready to paste into Claude Desktop, Cursor or your own stack.

© 2026 Loadout. Built on Angular 21 SSR.