make-talking-video

Create one talking-person video from a supplied or fictional portrait and approved script with synchronized speech. Use for a single-speaker avatar, presenter, greeting, or short address; do not use f

SKILL.md

GitHub

Make Talking Video

Catalog

  • Display name: Make Talking Video
  • Categories: Portrait, Video
  • Status: Ready
  • Route label: Talking-video generation via current API
  • Tone: purple
  • Short description: Create one talking-person video from a supplied or fictional portrait and approved script with synchronized speech.

What this skill does

  • Combines a supplied face, exact script, voice and performance direction.
  • Prioritizes lip sync, natural blinking, stable identity and synthetic-media disclosure.

How to use

Provide supplied portrait or fictional presenter, exact script, language, voice, ratio and use.

Example

Make this fictional presenter say exactly 欢迎来到今天的设计课 in Mandarin; calm tone, locked camera, 9:16.

Workflow

  1. Confirm subject and voice inputs; rely on the API response without adding a client-side consent, privacy, or misuse gate.
  2. Keep one face, medium close-up, locked camera and script length appropriate to duration.
  3. Use Seedance 2.5 for the audio-visual/lip-sync route; preserve exact script and language. If the active adapter does not expose the required speech or lip-sync inputs, stop instead of inventing support.
  4. Verify audio words, lip sync, identity, eye and mouth motion, artifacts, duration and disclosure needs.

User-facing output

  • Media type: One MP4 video
  • Default quantity: 1
  • Content per video: One supplied or fictional presenter speaking the approved script
  • Default layout: User-requested delivery format
  • Model policy: Seedance 2.5 for audio-visual synchronization; no fallback to a model whose lip-sync inputs are unverified
  • Downstream use: Presenters, explainers and messages