The Pulse
ElevenLabs puts 50-plus creative models behind one MCP connector
ElevenLabs has expanded its MCP connector beyond agent management to generate speech, transcripts, dubs, music, sound effects, images, and video. The company says the connector is available in Claude, ChatGPT, Cursor, Grokbot, Hermes, and o
AI.info Team ·
More than 50 models now sit behind a single ElevenLabs connector for AI assistants. The company announced on September 14 that its MCP integration can generate speech, transcripts, dubs, music, sound effects, images, and video, extending a tool that previously focused on managing ElevenLabs agents.
The expanded connector is available in Claude, ChatGPT, Cursor, Grokbot, Hermes, and other assistants, according to ElevenLabs’ announcement. Users install the connector, sign in through OAuth, and send creative requests from the assistant they already use. ElevenLabs says the setup requires no local server, API keys, or manual configuration.
One MCP connection now covers audio and video
The connector brings several parts of ElevenLabs’ product catalog into the same conversation. Text-to-speech can produce a voiceover using a library voice, while Scribe turns recordings into transcripts with speaker labels and timestamps across 99 languages. Dubbing can then create another-language version while preserving the original speaker’s voice, tone, and delivery.
Music and sound-effects generation add another layer to the workflow. Users can request a track in a specified genre or style, with or without vocals, and generate multiple sound-effect variations from a prompt. Those outputs can sit alongside narration and video rather than requiring separate handoffs between applications.
Images, animation, and lip-sync join the same thread
ElevenLabs also connects image and video generation to the audio tools. A user can generate or edit an image, animate it into video, and add lip-sync within the same assistant thread. The company describes the system as a way to move from a brief to a script, voiceover, music bed, sound design, and finished video without assembling each part in a separate application.
That workflow builds on ElevenCreative, the company’s broader creative platform. Every generation made through the MCP connector is sent to the user’s ElevenCreative workspace, where it can be opened in Studio for timeline edits, narration changes, music and sound-effects layering, and export.
ElevenLabs links creative tools to its agent infrastructure
The new capabilities use the same MCP connection that powers ElevenLabs Agents. Teams that already connected the service to review agent performance or create agents do not need a second installation for the creative tools, the company says.
ElevenLabs has also added workspace controls around the connection. Administrators can decide which tools the integration is allowed to call, while data residency is selected when the connection is created. Access is scoped to the user’s workspace through OAuth.
From local MCP server to hosted connector
The update marks a change from ElevenLabs’ original MCP implementation. The company’s earlier April 2025 MCP announcement described a local, open-source server that forwarded requests to ElevenLabs’ cloud APIs and required developers to manage an API key and local installation.
ElevenLabs later introduced a hosted MCP server, and its current product announcement presents the connector as a service that can be added through assistant connector directories. The company’s stated goal is to put the same creative functions available in ElevenCreative and Flows inside conversational tools, while keeping the resulting assets in the ElevenCreative workspace.
What the connector changes for creative teams
The most significant change is not a new generation model but the location of the workflow. A creator can request several linked assets from one conversation instead of moving a script, audio track, image, and video between unrelated tools. ElevenLabs says every node available in Flows is also available through the MCP connector, aligning the assistant-based workflow with the company’s visual creative canvas.
That arrangement still leaves final control in Studio rather than the assistant alone. The connector is designed to produce a first version and assemble the components; users can then adjust timing, narration, music, sound effects, and the final cut in ElevenCreative. Availability begins September 14 through the listed assistant integrations, with ElevenLabs directing users to its MCP page to connect an account.