Avatar script over 60 seconds: split it into two Avatar 1.0 jobs

Sume accepts avatar scripts that it estimates at 4 to 60 seconds. Longer scripts need to be shortened or split into jobs; here is how to split without a jump.

5 min readSume
All posts

A talking-avatar script on Sume has to land between 4 and 60 seconds of estimated video. Past 60 seconds the request is not accepted, so you either cut the script or split it into two or more jobs and join the results. This post shows where to cut and what the docs say about each limit.

The limits below come from the Generate avatar video docs (read 2026-10-06).

What is the length window?

POST /v1/avatar-1.0/talking-video takes either a script or ordered video_inputs, never both. The docs say Sume accepts scripts and multi-scene plans when it estimates the target duration at 4 to 60 seconds inclusive, and that longer scripts should be made shorter or split into multiple jobs. The same window applies to the total planned duration of a video_inputs plan.

Avatar video length rules (read 2026-10-06)
InputRuleSource page
script or video_inputs totalEstimated 4-60 seconds, inclusiveGenerate avatar video
Inline captionsRejected when the estimated duration is over 60 secondsGenerate avatar video
Avatars per final videoOne resolved avatarGenerate avatar video

Where should I split the script?

Split at a sentence end, where the speaker would naturally pause, and give each half its own job with its own Idempotency-Key. Each job is a separate talking video with the same avatar_handle, so the face and the voice stay the same. Keep the two halves under 60 seconds each so inline captions stay available on both.

  • Cut on a full stop, not in the middle of a thought, so the join does not land on a half-said sentence.
  • Reuse the same avatar_handle and the same scene direction in both requests. The docs say the current execution supports one resolved avatar for each final video and expects scene backgrounds to resolve to one shared scene.
  • Poll each job with GET /v1/jobs/{id}/status and read GET /v1/jobs/{id}/result; see Jobs and results.
  • Join the two MP4s afterward in a Timeline render; the Timeline docs list the clip slots and transitions.

Do I need a split at all?

If the script is 70 seconds because it repeats itself, cutting words is cheaper than a second job. If it really is two beats, such as a hook and a long demo, a multi-scene video_inputs plan can hold a voice.type: "silence" beat with a required duration, but the whole plan still has to fit in 60 seconds.

Sources

More in Sume Avatar 1.0

All Sume Avatar 1.0 posts

Written by Sume