Avatar video script too long? The 4 to 60 second rule

Keep each Avatar 1.0 video's script to a length Sume estimates at 4 to 60 seconds inclusive, or the request is rejected. The avatar videos docs say to shorten longer scripts or split them into separate videos. A script that is too short, under 4 seconds, is out of range too, so a one-word line needs padding or a longer beat.

The window applies to a plain script and to a multi-scene video_inputs plan alike: the total planned duration must land in the same range.

How do you estimate before you submit?

Sume does the estimate server-side, and the docs do not publish its formula, so do not rely on a word-count rule of your own for borderline scripts. A rough local check is still useful for catching the obvious cases. Spoken English commonly runs near 2 to 3 words per second, so a script of 150 words is already near the limit, and 200 words is probably over. Treat that as a screening heuristic and not a guarantee.

How should you split a long script?

For a plan that mixes spoken lines with pauses, video_inputs accepts a silence beat, voice.type: "silence", which needs a duration and no script. That is how you add a breath without adding words.

• Cut at sentence ends, so each video is a complete thought.
• Aim for 30 to 45 seconds per piece so a small estimate error does not push one over the line.
• Use the same avatar and settings on every piece for a consistent look.
• Number the pieces in your own system and join the finished videos in your editor or a timeline step.

Do captions change the limit?

Inline captions are opt-in and burn the style into the final MP4 after generation, using the spoken script. An estimated duration above 60 seconds is rejected for them. They also soft-fail: the avatar job can still succeed with a clean primary video and captions.status=failed. Inline captions do not create a separate billed caption job.

If you translate scripts first, budget length per language, as in the length-budget post, because a longer translation can push a script past 60 seconds. Check the current options in the avatar videos docs and the Avatar 1.0 docs.

What do you do about a script that sits near the limit?

Scripts that estimate at 58 or 59 seconds are risky, because the server's estimate can differ from your own. Trim them to leave a margin, or split them. A second reason to split is editing: shorter videos are easier to replace when one line changes, since you only regenerate the piece that changed and not the whole script.

Keep a record of each piece's script and its estimated duration, so a rejection is easy to diagnose.