Split a Short's audio at a cut point: timeline-audio ranges

If you want to split a sound and reuse the halves, POST /v1/timeline-1.0/audio takes a source and up to 20 ranges, or up to 20 parts to join, and returns a WAV for $0.01 a job. The join is done in the sample domain, so two parts butt together without the gap that comes from decoding and re-encoding.

YouTube's side: the audio page lists split, volume, fade, find beats, cleanup and delete (read 2026-10-03, Add & edit audio). Sume's side: the Timeline audio doc.

When is a split useful?

Typical needs in a Short: drop a cough in the middle of a voice take, move the punchline to the front, or keep only the first 12 seconds of a long music track.

• Concat mode joins parts in order, for example intro, take 3, outro.
• Split mode returns several clips from one file when you pass ranges.
• WAV is the default output; MP3 adds encoder priming padding, so avoid it for sample-exact seams.

How is it priced and polled?

It is a flat $0.01 job. As with other Sume jobs, send an Idempotency-Key; the default is async, and a sync call waits up to 30 seconds before it returns the job to poll via GET /v1/jobs/:id/status and then /result. Details are in the jobs doc.

How does it feed a render?

The returned file goes into Timeline as audio.url, with duration_seconds matching the new length. If you only want the parts joined inside a render and no separate file, Timeline accepts audio.parts[] directly, also up to 20 slices, exclusive with audio.url.

What Sume does not do

It does not find the cut point for you or detect beats; you supply the seconds. YouTube's find-beats and cleanup tools have no counterpart here.