Hook, demo and CTA in one avatar video with a silence beat

Send ordered video_inputs to POST /v1/avatar-1.0/talking-video instead of one script. Each entry is a beat with a voice: type: text for speech, or type: silence with a required duration for a pause. The total planned duration must be 4 to 60 seconds, and all scenes share one avatar and one scene.

The three beats

A short ad usually has a hook, a moment where the product is on screen, and a call to action. In video_inputs that is two spoken entries and one silent entry. A silent beat has no speech, so script and input_text are not permitted on it.

Spoken entries use type: text with one of script or input_text, never both. Give each entry an id and a duration in seconds, and a background such as a prompt.

curl -X POST https://api.sume.com/v1/avatar-1.0/talking-video \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: avatar-video-multi-001" \
  -d '{
    "avatar_handle": "sume_clawra",
    "aspect_ratio": "9:16",
    "quality": "plus",
    "video_inputs": [
      { "id": "hook", "voice": { "type": "text", "script": "Wait, this turned one selfie into a whole video?", "duration": 3 },
        "background": { "type": "prompt", "prompt": "Casual bedroom framing, native UGC lighting" } },
      { "id": "demo", "voice": { "type": "silence", "duration": 4 },
        "background": { "type": "prompt", "prompt": "Casual bedroom framing, native UGC lighting" } },
      { "id": "cta", "voice": { "type": "text", "input_text": "Pick a template, drop in your photo, and it builds the clip.", "duration": 5 },
        "background": { "type": "prompt", "prompt": "Casual bedroom framing, native UGC lighting" } }
    ]
  }'

The constraints

What the silent beat is for

The silent beat is where the product or scene carries the message. Pair it with product_image or a photo scene (scene: { "type": "photo", "image_url": "https://..." }), and keep it short. Four seconds of nothing said feels long in a 12-second video.

Media fields must be fetchable public HTTPS URLs. Sume rejects localhost, private-network and non-HTTPS addresses before it submits.

Review before you render

Run the same body as a preview first. You get one still per scene in scene_previews[]. Later scene stills are pose-anchored continuations of the first frame when the scene is shared, so you see the continuity before you pay for the video.

Add captions with enabled: true and a style if you want them burned into the final video. Korean scripts need a Hangul style.