Avatar clip under 4 seconds: add a silence beat or a longer line

Sume accepts avatar scripts and plans that it estimates at 4 to 60 seconds, so a one-line script that reads in 2 seconds is below the window. Two honest fixes are a longer line, or a multi-scene video_inputs plan that adds a voice.type: "silence" beat with a required duration.

Rules from the Generate avatar video docs (read 2026-10-06).

How does a silence beat work?

In video_inputs, each scene has a voice of type: "text" with a script or input_text, or type: "silence" with a required duration and no script. The total planned duration must still be 4 to 60 seconds. The docs example uses a 3-second hook, a 4-second silent demo and a 5-second call to action.

{
  "avatar_handle": "product_host",
  "aspect_ratio": "9:16",
  "video_inputs": [
    { "id": "hook", "voice": { "type": "text", "script": "New drop today.", "duration": 2 } },
    { "id": "beat", "voice": { "type": "silence", "duration": 3 } }
  ]
}

What about padding with words?

Filler words make a weaker clip than a pause. If the pause is a product shot, use a scene background that shows it. All scenes share one avatar and resolve to one shared scene in the current execution, so plan the beat around the same setting.

What does it cost?

Check the avatar-video estimate in the catalog (GET /v1/catalog) for your quality tier and length before you batch a series, since the price follows the planned seconds.