AI clip detection
The transcript is read for passages that open on a hook, make one point and resolve — then checked against the real video duration.
One long video in. Several vertical clips out — cut at the passages that stand on their own, reframed to 9:16, captioned in sync with the speech.
Upload and set up your first project without paying. Plans from €14.99 a month.
Made with ClipAspect
One long video. Several vertical clips.
How it works
Transcribed to the word, read for passages that stand on their own, cropped to 9:16 around the subject, captioned and encoded. None of it by hand.
passages kept, scored on how well they stand alone
This changed everything
Nobody tells you
Here is the proof
What it actually does
Finding the passage worth clipping, keeping the subject in a vertical frame, and landing captions on the syllable. Everything else is plumbing.
The transcript is read for passages that open on a hook, make one point and resolve — then checked against the real video duration.
Every passage gets a score. Low ones are meant to look low — padding the list would waste your allowance.
A timestamp on every word, so captions land on the syllable and the spoken word lights up as it is said.
We charged too
Frames are scored for edge detail and motion. The crop lands on the subject and holds steady — never stretched.
Several clips from one upload, spread across the video so they do not all come from the same five minutes.
One FFmpeg pass does the cut, the reframe and the caption burn-in, so the frame is only scaled once.
Caption styles
These previews are driven by the same configuration the renderer reads, so what you see here is what gets burned into the file.
You can change this
Clean
White text with a hard black outline. Readable on anything.
You can change this
Highlight
The word being spoken lights up as the audio plays.
You can change this
Bold Center
Large centred type. Works well for talking-head clips.
You can change this
Minimal
Small, quiet captions that stay out of the way.
You can change this
Creator
Punchy type with a coloured active word and a pop on entry.
Before / after
By hand
With ClipAspect
Who it is for
If you have footage and not enough hours, this is the part of the workflow it removes.
Turn a weekly episode into the clips that bring people to the full episode.
Pull Shorts out of long uploads without recutting them in an editor.
Find the moments in a multi-hour VOD without scrubbing through it.
Lift the explanation that actually lands out of a long lesson.
Get short-form out of the interviews and talks you already recorded.
Produce a week of vertical output from one recording session.
Pricing
Every plan includes every feature — clip detection, captions, reframing and export. What changes is how many videos you can put through each month. You see the exact cost before each one, and a failed job is refunded automatically.
€14.99/month
≈ 10–12videos / month
For getting your first clips out.
3,200 credits a month
€29.99/month
≈ 30–35videos / month
For creators publishing several times a week.
9,600 credits a month
€79.99/month
≈ 100–120videos / month
For high-volume creators, agencies and teams.
32,000 credits a month
Cancel any time. No refunds on time already paid for. No minimum term. Read the terms
FAQ
No tool can tell you that, and anything claiming otherwise is guessing. What this does is measurable: it finds passages that open on a hook, make one point and resolve, then scores them on those criteria. Whether a clip performs depends on your audience, your topic and timing.
MP4, MOV, M4V, WebM, MKV files that you own or have permission to use. You confirm that when you create a project. There is no download-from-a-link feature: this does not fetch content from YouTube or any other platform, and it will not work around a platform's restrictions.
Roughly a third of the video length for most files, dominated by transcription and by encoding each clip. You can close the tab — the work continues, and the project page shows where it is when you come back.
Yes. Correct a word, change how many words appear per line, move the captions, or switch style entirely. Only the affected clip is re-rendered; the transcript and the clip boundaries are untouched.
Short is 15–30s, Medium is 30–60s, Long is 60–90s. Medium suits most talking content. A clip is never cut mid-sentence to hit the number — the boundary moves to the nearest natural pause.
They are private to your account, served only through authenticated, short-lived links, and deleted along with the project whenever you delete it. Nothing is made public.
You get the specific reason — not a generic error — and the credits reserved for that job are returned to your balance automatically.
Upload it and see which passages come back.