Source
One file, length capped by your plan.
Video repurposing / transcript to clip
Turn one recording into clips worth publishing.
Upload a podcast, interview, webinar or long-form video. It gets transcribed with word-level timings, a model picks the moments worth cutting, and each one comes back as a vertical clip with optional captions burned in.
Outputs from one source video
Long recordings in. Ready-to-post clips out.
What is cutspool?
cutspool is a web app that turns one long recording into short vertical clips. It transcribes the source with word-level timings, has a model read that transcript and return the ranges worth cutting, then cuts each range, reframes it to vertical 9:16 and burns word-by-word captions into the picture. Each finished clip is an MP4 with H.264 video and AAC audio at 1080 x 1920 or 720 x 1280. A run costs one credit per minute of source video, rounded up.
The workflow
One timeline. Every useful moment.
Transcription, moment selection, trimming, framing, captions and export all read from the same uploaded file and the same word-level transcript. A clip is a range on that timeline, not a separate document, so the words, the cut and the caption timing can never drift apart.
18:24So we ran the seasonal discount for a third year…
18:48The discount that quietly trained your best customers to wait.
19:29…the same accounts bought at full price within a quarter.
Transcript
Every word, with a start and an end.
00:0018:48 – 19:2958:42
Range
Up to four moments make one clip.
Output
A vertical MP4 with captions burned in.
And that is the whole product
Every clip, from the same source.
Find the moments
Start with the moments that carry the conversation.
The transcript goes to a model that returns clip ranges rather than a summary. Each candidate comes back with a title, a short description, hashtags and a score from 0 to 10, and the list arrives sorted with the strongest first.
A single clip can stitch up to four separate moments from the same recording, so an answer that arrives in two passes still reads as one cut.
TranscribeFind the momentCutDeliver
Word-level transcript
So we ran the seasonal discount for a third year, and on the day the numbers looked fine.
But when we pulled the cohort out afterwards the picture changed completely.
The discount that quietly trained your best customers to wait for the next offer instead of buying today.
Once we stopped running it, the same accounts bought at full price within a quarter.
Before the run
Set the length. The cut lands on a sentence.
Before a run you choose the clip length. The AI returns every strong qualifying moment within your plan limit. Word-level timings mean the boundaries land on sentence edges instead of mid-breath.
Pause removal drops silences over 1.5, 2 or 3 seconds; tightening removes long pauses and hesitation sounds, then speeds delivery up by 8%. The model chooses the original ranges. While the source is retained, each generated clip can be shortened or re-extended inside its original timeline.
Framing
Frame once. Export for the feed.
A wide source frame becomes 1080 x 1920 or 720 x 1280. The default pads the frame with a blurred copy of itself so nothing at the edges is lost; fill-frame crops in instead, and you can slide that crop until the speaker sits where you want them - before the run or afterwards while the source is retained. Whichever you pick holds for the whole clip and does not follow the speaker.
Captions
Captions timed to the words.
Captions are built from the same word timings that place the cut and are burned into the file, so they survive re-uploads and platform re-encodes. The word being spoken flips to a highlight colour. 12 presets are the starting points, and every field behind them is yours to set.
White words, yellow on the beat, hard black outline.
- Font
- Open Sans
- Words per line
- 3
- Highlight
- #FFE500
- Behind the words
- 4px outline
- Entrance
- None
Source to MP4, five stages
Watch a run go through.
One file goes in.
Upload a podcast, interview, webinar or long-form recording. Credits are charged on the length of the file you upload, and returned if the run fails or you cancel it.
webinar-migrations-q3.mp4
Use cases
One long recording, many short ones.
What it takes and what it gives back.
These are the limits the product enforces today. Anything not listed here is not implemented yet.
- Supported input
- MP4, MOV, WebM
- Maximum file size
- 2 GB to 8 GB, by plan
- Maximum source length
- 60 minutes to 4 hours, by plan
- Clips per run
- 2 to 15, by plan
- Clip length
- 10 to 180 seconds, set before the run
- Output
- 1080 x 1920 or 720 x 1280, 9:16
- Output format
- MP4, H.264 video, AAC audio
- Captions
- Burned in, word-timed, 12 presets
- Usage unit
- 1 credit per rounded-up source minute
Pricing
Pay for the footage you process.
See full pricingOne credit covers one source minute, rounded up to the next whole minute. Credits are returned if a run fails or is cancelled.
Creator
€9.99
Occasional creators. 150 credits a month, about 150 minutes of source video.
Pro
€29.99
Regular creators and podcasters. 750 credits a month, about 750 minutes of source video.
Studio
€59.99
Heavy users and agencies. 2000 credits a month, about 2000 minutes of source video.
Enterprise
Contact us
Teams and high-volume workflows · Custom monthly credits · API access · highest queue priority or dedicated capacity · custom retention and upload limits · centralized billing or invoices · priority support · optional onboarding · volume pricing.
Before you upload.
MP4, MOV and WebM. The size and length ceilings are set by your plan, from 2 GB and 60 minutes at the lowest to 8 GB and four hours at the highest. The pricing page lists both per plan. A file past your plan's limit is rejected before anything is charged.
The recording is transcribed with word-level timings, then a model reads the transcript and returns clip ranges. Each clip comes back with a title, a description, hashtags and a score from 0 to 10. A clip can stitch up to four separate moments from the same recording.
Not arbitrarily. You set the clip length before the run, and the model returns every strong qualifying moment up to your plan's limit. Once a generated clip exists, you can shorten it or re-extend it only inside its original generated timeline while the source is retained with READY status. Picking a new moment outside that timeline requires another run; once the source expires or is deleted, source-backed editing is unavailable.
Yes, while the source is retained with READY status. The editor shows the clip's own transcript on its timeline, and any word can be corrected - useful when a name or a technical term came back wrong. A correction changes the text that is burned in; it does not choose a new source moment.
Yes, before the run and afterwards while the source is retained with READY status. The default pads the frame with a blurred copy of itself so nothing is cut off; fill-frame crops in instead, and you can slide that crop left or right until the speaker sits where you want them. Whatever you pick holds for the whole clip: it does not step, track the speaker, or cut to B-roll.
Yes, while the source is retained with READY status. Start from one of 12 named presets, then set the font from 7 faces, the size, the weight, uppercase, the base and highlight colours, the outline colour and width, the shadow, a solid background box, the vertical position and the words per line. A style carries an outline or a background box, never both, because the renderer's box replaces the outline instead of sitting behind it. Free plans get 2 presets and their highlight colour; paid plans get all 12 and every field.
No. The guides are preview-only overlays for checking where each platform may cover the picture with its interface. They are never burned into the rendered clip.
MP4, H.264 video and AAC audio, at 1080 x 1920 or 720 x 1280. Paid plans pick between the two and default to 1080 x 1920; Free renders at 720 x 1280. 9:16 is the only output ratio.
One credit per source minute, rounded up to the next whole minute. You are charged on the file length, not the number of clips. If a run fails, the credits are returned.
The job is retried automatically. If it fails for good, the credits are refunded and the partial output is cleaned up. You can also cancel a queued job and get the credits back.
Ready when the recording is