ElevenLabs MCP

ElevenLabs MCP is the Model Context Protocol server for the ElevenLabs voice platform. It exposes speech generation and audio processing as callable tools, so an assistant can produce narration, transcribe a recording, or build a voice asset as a step inside a larger task rather than as a manual detour into a separate web app. The main surface is text to speech: an agent picks a voice from the account or the shared voice library, sends text, and gets an audio file written to disk. Around that sit the rest of the platform's audio functions. Speech to text turns uploaded recordings into transcripts with speaker separation. Voice cloning creates a new voice from sample audio, and voice design generates one from a written description. Sound effect generation produces short non-speech audio from a prompt, and audio isolation strips background noise out of a recording. The server also reaches the conversational agent side of the account, so voice agents can be listed, created, and configured, and outbound phone calls placed through a connected number. It suits developers wiring voice into a product, content teams producing narration and dubbing at volume, and anyone who wants a coding assistant to generate and iterate on audio files locally. Because generation consumes account credits and cloning touches voice identity, most setups keep the server on a scoped API key and review which tools it is allowed to call.