Cut a 10-minute vertical video into Shorts episodes at sentence ends

A 10-minute vertical video needs at least four YouTube Shorts episodes, because each episode can be at most 180 seconds (600 divided by 180 is 3.33). Use Sume video inspect with segmentation.mode: "sentence" to find where sentences end, then cut four ranges with video trim: about $0.10 for the transcript and $0.08 for four trims.

TikTok lets you record up to 10 minutes in the app, per its Camera tools support page, so a 10-minute master is a normal TikTok length. YouTube's Shorts series rule is portrait or square, 180 seconds at most, and the playlist must contain only Shorts (all read 2026-10-06).

How do I find cut points that do not split a sentence?

Send POST /v1/video-inspect with transcribe: true, segmentation: {"mode": "sentence"}, duration_seconds: 600, and frames: false. Sume returns sentence segments[] with no gaps between them. The duration_seconds hint is capped at 600 seconds, which is exactly your 10 minutes, and the transcript is billed at $0.01 per audio minute, so $0.10.

Walk the segments and start a new episode at the last sentence end before the 180-second mark. For a steady talk this gives four episodes of about 150 seconds, which leaves room to spare under the cap.

How do I cut the episodes?

For each range, call POST /v1/video-trim with video_url, start, and duration (not end too; sending both is video_trim_range_conflict). Leave precision at exact so the cut is frame-accurate. Each trim is $0.02 and returns a new MP4 in video_url. Source length may be up to 1800 seconds and output up to 900, so a 10-minute master is within both.

What does the whole split cost?

Rates come from the video inspect and video trim docs. Confirm the live prices in GET /v1/catalog. The probe-and-frames part of inspect is metered separately, so check its price as well.