Library song capped at 90 s in a 3-minute Short: 90 seconds left bare

The rule and the gap

YouTube's Shorts help page (read 2026-10-09) says most songs from the Shorts Audio Library are limited to "up to 90 seconds in a 3 minute Short", and some tracks have stricter 60 or 30-second caps. In a full 180-second Short that leaves 90 seconds with no library track at the common cap, 120 seconds at the 60-second cap and 150 seconds at the 30-second cap.

Three ways to fill it

You can leave the rest as voice only, you can build the entire soundtrack yourself so no library track is involved, or you can split the idea into shorter Shorts. The page also says a Short over one minute with an active copyright claim of any type is blocked globally and cannot be monetized, so a bed you did not license is a larger risk in a long Short than in a 45-second one.

Scoring the whole 180 seconds on Sume

Sume's Timeline 1.0 takes one audio spine up to 1,800 seconds plus an optional soundtrack bed with loop, fade_out_seconds up to 10 and duck_db from 0 to 20 so the bed sits under speech. Music 1.0 is a flat-priced generation: the docs say not to send a duration, and the catalog lists $0.125 per generation (read 2026-10-08). Because the bed's length is not chosen, the loop flag is the way to cover a longer Short.

• Render 180 s: ceil(3) x $0.10 = $0.30.
• One Music 1.0 bed: $0.125.
• Total for the bed and render: $0.30 + $0.125 = $0.425 (voiceover and clip costs excluded).

What not to assume

A generated bed is not automatically free of claims, and Sume's docs do not make a claims-clearance promise, so check the result in YouTube's own copyright flow before relying on monetization. The library cap applies only to tracks picked inside YouTube's tools; a file you upload with its own soundtrack is not subject to that cap, though the claim rule above still is.

A worked 180-second plan

Suppose the Short is 180 seconds and you pick a library song with the 90-second cap. The song covers the first half only. A voice-led Short can simply run the second 90 seconds on voice and room tone, and many do. A Short that needs continuous music can instead carry its own soundtrack end to end.

In the Sume version, the audio spine is the voiceover (or a declared silent spine), and the soundtrack bed sits under it with duck_db set so the music dips when speech starts. A 180-second Timeline job is 3 billed minutes, $0.30, and the bed is one flat $0.125 generation. If the finished file comes out longer than 3 minutes it is no longer a Short under the page's rule, so set audio.duration_seconds to 180 or less.