Repurpose YouTube Videos into Shorts with Better Captions
Reuse the long video’s caption file instead of re-transcribing: keep the cues in range, shift them by the clip start offset, then re-wrap for the vertical frame.
Turning a long YouTube video into Shorts sounds simple until the captions get involved. A ten-minute upload might contain several good clips, and the usual instinct is to send each one back through transcription as if it were new footage. It is not new footage. The long video already has a caption file, and most of the work is reusing it correctly.
Reuse the caption file you already have
If the source video is captioned, the words and the timings for your clip already exist. Re-transcribing throws that away, along with every spelling correction anyone made. The reuse is three mechanical steps:
- Export the long video's captions as SRT or VTT.
- Keep only the cues that fall inside the clip you are cutting.
- Shift those cues back so the first one starts near zero, by the same offset as the clip start.
The third step is the one people get wrong, because the timestamps are still measured from the start of the ten-minute video. If your clip begins at 4:12, every cue is 252 seconds too late:
clip starts at 4:12 in the source = 252 seconds
00:04:15,000 --> 00:04:17,400 cue in the long video
00:00:03,000 --> 00:00:05,400 same cue in the ShortApplying one offset across every cue is what the subtitle timing shifter does, so this does not need to be done by hand. Transcribe from scratch only when the source has no caption track, or when the clip audio was re-edited.
Re-wrap for the vertical frame
Reused captions still carry the line lengths they were given for a wide frame. Broadcast practice puts that near 42 characters a line, and a vertical frame has less usable width once the platform interface is drawn over it, so short-form wants closer to 32. A line that was comfortable in the long video wraps somewhere you did not choose in the Short. Where the break should land, and why an even split is the wrong goal, is covered in caption line breaks for short-form video.
Positioning is a format decision
The Shorts interface covers the bottom of the picture, which is exactly where a caption defaults to. Whether you can do anything about that in the file depends on which format you kept: SRT cannot say where a caption goes at all, and WebVTT can. If placement matters, carry the clip as VTT rather than SRT, or move the caption in the editor before burning it in.
Pick the clip on one idea, not one length
A useful Short usually carries one clear idea rather than a compressed summary of the whole video. Choose the moment first and let its length follow, then check the reused captions against it: a clip that starts mid-sentence inherits a mid-sentence first cue, which reads as a mistake even though the timing is correct. Trim the opening cue or re-record the hook.
Review what the reuse cannot fix
Reused captions inherit the long video's decisions, including the ones that were wrong. Names, product terms, numbers, and code-switched lines are worth reading once more, because an error copied from the source is still an error in the Short, and it now sits on a clip that may be seen far more often than the original.
The workflow, end to end
- Cut the long video down to one clear vertical clip.
- Export the source captions, keep the cues in range, and shift them by the clip start offset.
- Re-wrap the lines for a vertical frame, nearer 32 characters than 42.
- Keep VTT if caption placement matters, SRT if it does not.
- Read the names, numbers, and first cue once before publishing.
AutoCaption covers the caption half of that in the browser. Upload the clipped video in the caption generator, review the timed captions, and export the subtitle or styled video file your channel needs. For the Shorts workflow on its own, the YouTube Shorts captions page covers captioning the final vertical clip end to end.
Quick answers
Do I need to re-transcribe a clip taken from a captioned video?
Usually not. The words and timings already exist in the source caption file, along with any corrections that were made to it. Keep the cues that fall inside the clip and shift them back by the clip start offset. Transcribe again only when the source has no caption track or the clip audio was re-edited.
Why are my reused captions showing at the wrong time?
Because the timestamps are still measured from the start of the long video. A clip beginning at 4:12 carries cues that are 252 seconds too late, so every cue needs the same offset subtracted before the file works against the Short.
What has to change when a caption moves to a vertical frame?
Line width and placement. Lines written near the broadcast figure of 42 characters want to be closer to 32 in a vertical frame, and the Shorts interface covers the area a caption defaults to. SRT cannot carry position information, so keep the clip as WebVTT if placement matters.