The outcome

Create a short synthetic speaking performance with accurate words, identity approval, disclosure, and safe distribution.

Step by step

A workflow you can repeat.

  1. 01

    Document consent for the face, voice, exact script, language, audience, channels, commercial use, editing, retention, and revocation before uploading anything.

  2. 02

    Prepare a well-lit authorized portrait and clean audio, removing sensitive metadata, background voices, private content, copyrighted music, and unnecessary duration.

  3. 03

    Select lip-sync and the intended voice source, verify whether custom voice data is retained or shared, and generate a short low-cost test first.

  4. 04

    Review every word, pronunciation, lip timing, teeth, tongue, expression, identity, background, audio level, captions, and whether the clip implies an unapproved endorsement.

  5. 05

    Obtain final subject approval, preserve AI labels, add accessible captions and disclosure, publish only to the agreed audience, and remove unused identity assets when possible.

Working standard

What good use looks like.

  • Approve the exact script in advance.
  • Test with minimum audio.
  • Never imply an unauthorized endorsement.

Official references

Check the current product documentation.