ElevenLabs MCP server
ElevenLabs' official MCP server: text-to-speech, voice cloning, speech-to-text, sound effects, and conversational AI agents from your editor.
The ElevenLabs MCP server is ElevenLabs' official integration that exposes its audio and voice-AI platform as agent tools. From a chat or coding agent you can synthesize speech in hundreds of voices, transcribe audio with speaker diarization, clone a voice from sample clips, design brand-new voices from a text prompt, generate sound effects, isolate a voice from background noise, and convert audio from one voice to another. Beyond raw audio it also drives ElevenLabs' Conversational AI: the agent can create and configure voice agents, attach knowledge-base documents, list agents, and read back full conversation transcripts.
The server is published as the elevenlabs-mcp Python package and is meant to be launched with uvx (uvx elevenlabs-mcp) or installed via pip and run as python -m elevenlabs_mcp. It authenticates with an ELEVENLABS_API_KEY from your account settings; the free tier includes a monthly credit allowance. Optional configuration controls where generated files are written (ELEVENLABS_MCP_BASE_PATH), how results are returned to the client (as files, MCP resources, or both), and the data-residency region for enterprise deployments. Generated audio is saved to disk and surfaced back to the agent so it can play, attach, or post-process the output.
Quick install
Copy-paste configs are provided for all 8 supported clients. Pick your client below.
Add to ~/.claude.json
{
"mcpServers": {
"elevenlabs": {
"command": "uvx",
"args": [
"elevenlabs-mcp"
],
"env": {
"ELEVENLABS_API_KEY": "<ELEVENLABS_API_KEY>",
"ELEVENLABS_MCP_BASE_PATH": "<ELEVENLABS_MCP_BASE_PATH>"
}
}
}
}claude mcp add elevenlabs -- uvx elevenlabs-mcpAvailable tools
| Tool | Description |
|---|---|
| text_to_speech | Converts text to speech audio using a specified voice and model. |
| speech_to_text | Transcribes speech from an audio file, with optional speaker diarization. |
| text_to_sound_effects | Generates sound effects from a text description within a given duration. |
| search_voices | Searches existing voices by name, description, labels, or category. |
| list_models | Lists all available speech-synthesis models. |
| get_voice | Retrieves detailed information about a specific voice. |
| voice_clone | Creates an instant voice clone from provided audio sample files. |
| isolate_audio | Isolates the voice in an audio file by removing background noise and music. |
| check_subscription | Checks the current subscription status and API usage metrics. |
| speech_to_speech | Transforms audio from one voice into another while preserving delivery. |
| text_to_voice | Creates three voice preview variations from a text description. |
| add_generated_voice_to_library | Adds a generated voice preview to your voice library for reuse. |
| create_agent | Creates a Conversational AI agent with a custom configuration. |
| add_knowledge_base_to_agent | Adds knowledge-base documents to an agent via URL, file, or text. |
| list_agents | Lists all available Conversational AI agents. |
| get_agent | Retrieves detailed information about a specific agent. |
| list_conversations | Lists an agent's conversations with filtering and pagination options. |
| get_conversation | Fetches a conversation's details and full transcript by ID. |
Required configuration
- ELEVENLABS_API_KEYRequired
ElevenLabs API key from elevenlabs.io/app/settings/api-keys. Required.
- ELEVENLABS_MCP_BASE_PATHOptional
Base directory for reading and writing audio files. Optional; defaults to the user's Desktop.
- ELEVENLABS_MCP_OUTPUT_MODEOptional
How results are returned: files, resources, or both. Optional.
- ELEVENLABS_API_RESIDENCYOptional
Data-residency region (enterprise). Optional; defaults to us.
What you can do with it
Narrate and voice content in the agent loop
Generate voiceover, audiobook narration, or UI prompts with text_to_speech, design a bespoke brand voice with text_to_voice, and add sound effects with text_to_sound_effects, all without leaving your editor.
Transcribe and repurpose existing audio
Run speech_to_text with speaker diarization to turn recordings into searchable transcripts, isolate a clean voice track with isolate_audio, or restyle a clip into another voice with speech_to_speech.
FAQ
- Is it free?
- The MCP server is open source and free to run, and ElevenLabs offers a free tier with a monthly credit allowance. Beyond that, speech generation, transcription, and voice features consume credits billed against your ElevenLabs plan and API key.
- Does it support remote/OAuth?
- No. The server runs locally over stdio (via uvx or pip) and authenticates with your ELEVENLABS_API_KEY. There is no hosted remote endpoint or OAuth flow.