Skip to navigation
Plugins & ExtensionsAgent Skills

talking-photo skill

Make a photo, portrait or character talk.

Install this skill with npx skills add synchronicity-labs/skills --skill talking-photo, or get every skill with the Claude Code plugin. See Agent skills.

What it does

talking-photo turns a still image into a talking video. Give it a photo, a headshot or an illustrated character, plus either a script or a recording, and it returns an MP4 of that face speaking. The video is as long as the speech.

If you don’t have a picture yet and your agent can generate images, it offers to make one that works well: one clear face, looking at the camera, head and shoulders.

When to reach for it

What you haveReach for
A photo, portrait or illustrated charactertalking-photo
A video of a real personlipsync
A generated video clipai-video-to-talking

Try

  • “Make our CEO headshot say: welcome to Acme, we’re glad you’re here”
  • “Create a friendly cartoon barista and make her say good morning”
  • “Make ./portrait.png say what’s in ./greeting.mp3”

What you need

  • The sync. MCP server connected, or SYNC_API_KEY set. The setup skill walks you through either.
  • An image with one clear human or human-like face. Animals and non-humanoid characters aren’t supported.
  • A script, or an audio file, URL or sync. asset.

Common questions

Yes. It lists the voices on your account and suggests two or three that fit the character. Say “pick one yourself” and it chooses.

For a script, the length of the speech is estimated from the word count until it’s spoken. Speech synthesis is billed separately from the lipsync, and the agent says so.

sync. needs one clear face at a reasonable size. Group photos, tiny faces and side profiles are the usual causes. The agent explains which, and suggests a tighter head-and-shoulders crop.

No. The skill declines to put words in a real, identifiable person’s mouth to deceive. Your own photos, your team (with their OK) and fictional characters are fine.

It’s working if

  • The agent picks a voice from your account’s list, not a made-up ID.
  • You see the quote before the video is created.
  • The output is one MP4 of the face speaking your exact words.