Skip to main content
Glama
marzvfx

lipdub-mcp

Official
by marzvfx
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

Create LipDub render (uses account credits)

lipdub_create_render

Create a lip-sync render from a video URL and audio URL, making the speaker appear to say the audio. Confirm credit spend first, then get a render_id for async processing.

Instructions

Start a LipDub 2 render: make the person in a source video appear to speak a different audio track, with matched lip movement.

USES CREDITS. Each render charges the account and cannot be undone or refunded. Cost scales with the length of the source video. Confirm with the user first, then call again with confirm_spend set to true.

Supply both sides as direct, publicly downloadable URLs:

  • video_url — a video of ONE person speaking, face clearly visible.

  • audio_url — the audio you want them to appear to say.

LipDub 2 does NOT translate, transcribe or generate speech. If the user wants another language, produce that audio first with a separate text-to-speech or voice tool, host it at a URL, and pass that URL as audio_url.

Links must return the media file itself, with two exceptions: YouTube links work (youtube.com, youtu.be or youtube-nocookie.com), because they are resolved for you. Google Drive and Dropbox share pages, and anything behind a login, will fail.

Returns immediately with a render_id — the render is NOT finished. Rendering takes several minutes for a short clip, and longer for longer videos. Next step: call lipdub_wait_for_render with that render_id.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
audio_idNoAdvanced. A LipDub audio upload id from a previous upload, instead of audio_url.
video_idNoAdvanced. A LipDub shot id from a previous upload, instead of video_url.
audio_urlNoDirect URL to the audio the person should appear to say (.mp3, .wav, .m4a, .aac, .ogg, .flac, .mp4 or .mov). You must supply finished audio; LipDub 2 does not generate or translate speech.
video_urlNoURL to the source video (one person, face visible). .mp4, .mov or .avi, or a YouTube link.
callback_urlNoOptional https URL to be called when the render finishes. For automated pipelines; in a chat session use lipdub_wait_for_render instead.
confirm_spendNoWhether the user has personally approved spending credits on this render. Defaults to false deliberately, as a safety gate — it is not a misconfiguration. Set it to true only after you have asked the user and they have agreed. Setting it yourself without asking spends their money without consent.
output_filenameNoOptional name for the finished file, e.g. 'ceo-welcome-spanish.mp4'. If omitted, a name is derived from video_url. This name is also shown as the project name in the user's LipDub web app, so make it meaningful.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed3 schema fields changedv0.1.1
    • changedInput schema / properties / audio_url / description
      Previous value: -"Direct https URL to the audio the person should appear to say. You must supply finished audio; LipDub 2 does not generate or translate speech."New value: +"Direct URL to the audio the person should appear to say (.mp3, .wav, .m4a, .aac, .ogg, .flac, .mp4 or .mov). You must supply finished audio; LipDub 2 does not generate or translate speech."
    • removedInput schema / properties / transcript
      Removed value: -{
-  "description": "Optional. The words spoken in audio_url. Supplying this can improve lip-sync accuracy.",
-  "maxLength": 4096,
-  "type": "string"
-}
    • changedInput schema / properties / video_url / description
      Previous value: -"Direct https URL to the source video (one person, face visible). .mp4, .mov, .avi or .webm."New value: +"URL to the source video (one person, face visible). .mp4, .mov or .avi, or a YouTube link."
  2. First observedv0.1.0

TDQS

A5/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Goes well beyond the annotations by disclosing financial consequences ('USES CREDITS', 'cannot be undone or refunded'), cost scaling with source video length, asynchronous behavior (returns render_id immediately, render not finished), and typical render duration. No contradiction with the annotations; this description adds substantial behavioral context.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is front-loaded with the most critical information (purpose and cost warning), then organized into clear paragraphs and bullet lists covering inputs, exclusions, and post-call behavior. Every sentence earns its place; the length is justified by the tool's high-stakes, asynchronous, and easily-misused nature.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a tool with no output schema, the description fully explains what the caller should expect: an immediate render_id, that rendering is asynchronous, how long it may take, and the required next step. It also covers prerequisites, URL constraints, known failure modes, and the spending-confirmation gate, making it complete for correct invocation despite the schema listing all parameters as optional.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Even though schema coverage is 100%, the description adds crucial meaning: URLs must be direct, publicly downloadable media files; YouTube links are resolved but Google Drive/Dropbox share pages and login-protected links fail; video must show one person with a clear face; audio must be finished because LipDub does not generate speech. This meaningfully enriches the schema's parameter descriptions.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb ('Start a LipDub 2 render') and a clear resource with the intended effect: making the person in a source video appear to speak a different audio track with matched lip movement. This clearly distinguishes it from siblings like lipdub_list_renders, lipdub_get_render, and lipdub_wait_for_render, none of which create a render.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Provides explicit when-to-use and when-not-to-use guidance: confirms credits must be approved by the user first, instructs to call again with confirm_spend=true, warns that LipDub 2 does not translate/transcribe/generate speech, and routes follow-up to lipdub_wait_for_render. It also explains when callback_url is appropriate versus chat-session use.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools