Skip to main content

Digital Human

Create talking-head videos from a static portrait image. The AI generates lip-synced speech from your text and audio.

How to Use​

  1. Select a version/model:
    • LTX2.3 Digital Human (default) — Good quality, fast
    • wan2.2 Digital Human — Alternative model
    • MiniMax H3 Digital Human — High quality
  2. Upload a portrait image — A clear, front-facing photo works best
  3. Upload reference audio — A recording of the voice you want to clone
  4. Type the narration text — What the person should say
  5. Choose aspect ratio and duration
  6. Click Generate

Input Requirements​

  • Portrait Image — Clear, front-facing, well-lit, with the subject looking at the camera
  • Reference Audio — A clean voice recording (max 20 seconds) for voice cloning
  • Narration Text — Up to 1000 characters

Tips​

  • Use high-quality, well-lit portrait photos
  • Front-facing poses with a neutral expression work best
  • Audio files should be clean recordings without background noise
  • Keep reference audio under 20 seconds

Settings​

  • Aspect Ratio — 9:16 (portrait), 16:9 (landscape), 1:1 (square)
  • Duration — 4-10 seconds depending on the model
  • Max Edge — Resolution control (720, 1280, or 1920 pixels)