ElevenLabs MCP Server (Official)
OfficialOfficial ElevenLabs server for text-to-speech, voice cloning, sound effects, and audio.
Add to your client
Copy the config for your MCP client and paste it into its config file.
uvx elevenlabs-mcpPaste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"elevenlabs-mcp-server-official": {
"command": "uvx",
"args": [
"elevenlabs-mcp"
],
"env": {
"ELEVENLABS_API_KEY": "<your-api-key>"
}
}
}
}Requires `uv` (the Python package runner). Install it from https://docs.astral.sh/uv/ if `uvx` is not found.
Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf
Before you start
- An ElevenLabs API key from https://elevenlabs.io/app/settings/api-keys (free tier includes ~10,000 credits/month).
- Python with the `uv` package manager (or install via `pip install elevenlabs-mcp`).
- An MCP-compatible client (Claude Desktop, Cursor, Windsurf, OpenAI Agents). Windows users must enable Developer Mode in Claude Desktop.
- Optional: a microphone/audio setup if you want play_audio; ELEVENLABS_MCP_BASE_PATH to control where generated files land.
About ElevenLabs MCP Server (Official)
The official ElevenLabs MCP server gives AI agents access to the ElevenLabs audio platform — text-to-speech, voice cloning and design, sound effects, music generation, speech-to-text transcription, audio isolation, and Conversational AI agents. It lets a model turn text into spoken audio, transcribe recordings, design new voices, and even place outbound phone calls, all from inside an MCP client.
It's a Python package installed with uv (or pip as elevenlabs-mcp) and runs locally over stdio. You authenticate with an ElevenLabs API key. Generated audio can be written to files (default output to your Desktop), returned as MCP resources, or both, controlled via environment variables.
Beyond core TTS, it covers the broader ElevenLabs product surface: compose_music and create_composition_plan for music, create_agent / add_knowledge_base_to_agent and conversation tools for Conversational AI, and make_outbound_call / list_phone_numbers for telephony via Twilio or SIP.
It works with Claude Desktop, Cursor, Windsurf, and OpenAI Agents, and is maintained by ElevenLabs.
Tools & capabilities (14)
text_to_speechConvert text to spoken audio using a chosen voice and model.
speech_to_textTranscribe an audio file, with optional speaker diarization.
text_to_sound_effectsGenerate sound effects from a text description.
speech_to_speechTransform existing audio to a different target voice.
text_to_voiceGenerate three preview voices from a text description.
create_voice_from_previewSave a generated voice preview into your voice library.
voice_cloneCreate an instant voice clone from provided audio samples.
search_voicesFind voices by name, description, labels, or category.
search_voice_libraryBrowse the shared ElevenLabs community voice library.
isolate_audioExtract clean speech from audio with background noise.
compose_musicGenerate music from a prompt or a composition plan.
create_composition_planBuild a reusable structured plan for music generation.
create_agentCreate a Conversational AI agent with custom configuration.
make_outbound_callPlace an outbound call via Twilio or a SIP trunk.
When to use it
- Use it when you want your agent to narrate text or generate voiceovers with high-quality TTS.
- Use it when you need to transcribe audio or video, optionally separating speakers.
- Use it when you want to clone a voice or design a brand-new voice from a description.
- Use it when you need sound effects or generated music for a project.
- Use it when you're building or managing ElevenLabs Conversational AI agents and their knowledge bases.
- Use it when you want an agent to place outbound phone calls via Twilio or SIP.
Quick setup
- 1Get an API key at https://elevenlabs.io/app/settings/api-keys.
- 2Install with `uv` (recommended) or `pip install elevenlabs-mcp`.
- 3Run `python -m elevenlabs_mcp --api-key=YOUR_KEY --print` to generate client config, or add the server block to your client config manually.
- 4Set ELEVENLABS_API_KEY (and optionally ELEVENLABS_MCP_BASE_PATH / OUTPUT_MODE) for the server.
- 5Restart your MCP client (Claude Desktop, Cursor, etc.) so it loads the tools.
- 6Verify by calling check_subscription or text_to_speech and confirming an audio file is produced.
Security notes
Your ElevenLabs API key can incur usage charges and access your voice library; keep it in env, not in shared configs. The server can write generated audio files to a configurable local path.
ElevenLabs MCP Server (Official) FAQ
Where do I get the API key and is there a free tier?
Generate it at elevenlabs.io/app/settings/api-keys. The free tier includes roughly 10,000 credits per month, which is enough to test text-to-speech and other tools.
How do I install it — uv or pip?
Both work. `uv` is the recommended path for Claude Desktop; alternatively run `pip install elevenlabs-mcp`. It's a local stdio server, not a hosted remote one.
Where does generated audio go?
By default files are written to your Desktop. Control the location with ELEVENLABS_MCP_BASE_PATH and choose files/resources/both via ELEVENLABS_MCP_OUTPUT_MODE.
Which clients are supported?
It works with Claude Desktop, Cursor, Windsurf, and OpenAI Agents. On Windows, enable Developer Mode in Claude Desktop for it to load.
Can it do more than text-to-speech?
Yes — it also covers speech-to-text, voice cloning/design, sound effects, music generation, audio isolation, Conversational AI agents, and outbound phone calls.
Alternatives to ElevenLabs MCP Server (Official)
Compare all alternatives →
Popular community server that feeds Figma layout data to coding agents via a Figma API token.
Official MiniMax server for TTS, voice cloning, music, image, and video generation.
Official Figma server that brings design context, variables, and components into AI coding tools.
Compare ElevenLabs MCP Server (Official) with: