ElevenLabs MCP ServerUnclaimed

The ElevenLabs MCP server is the official Model Context Protocol server for ElevenLabs' text to speech and audio processing APIs. It lets MCP clients such as Claude Desktop, Cursor, Windsurf and OpenAI Agents generate speech, clone voices, transcribe audio and use the other audio tools the server exposes.

It is for people who already work in an MCP client and want audio tasks done from inside it, and for developers who want to run or change the server from source. Example requests on the page include creating an AI agent that speaks like a film noir detective, generating voice variations for a character, and turning a transcript back into speech with a different voice for each speaker.

Setup needs an ElevenLabs API key. For Claude Desktop, you add a short entry to claude_desktop_config.json that runs the server with uvx. Other clients can be set up by installing the elevenlabs-mcp package with pip and printing the configuration with a command. Generated files can be saved to disk, returned as MCP resources, or both. File access is limited to a base directory that you set.

The server uses ElevenLabs credits, so each tool call draws on your account. The page says there is a free tier with 10k credits per month. It does not list paid plan prices. Data residency is described as an enterprise feature.

The repository is archived and read-only. The page says the local server is deprecated in favour of ElevenLabs' hosted MCP server, which authenticates with OAuth and needs nothing installed locally. The code is released under the MIT license.