Recaption vs. transcription services
Captioning an old Vimeo library ends in one of two places: re-upload and let Vimeo caption the copies for free, or pay a service per minute of audio and attach the files yourself. Here's what each route actually costs — and where each one wins.
The two routes
Route one: transcription. Run every video through a speech-to-text service — Rev, Deepgram, or Vimeo's own Rev integration — get a caption file per video, and attach each file to the existing video. Views and comments are untouched, and human services hit 99% accuracy.
Route two: re-upload. Vimeo auto-captions every video uploaded after May 25, 2022 (paid plans, up to 8 hours), so uploading each old video again gets you captions with no transcription bill. Recaption automates route two across a whole library; transcription services are route one.
What 100 hours of video costs
Prices as published in July 2026 — check current rates before you budget. The shape of the math matters more than the decimals:
- Recaption: $29, once. Vimeo's auto captions are free; the fee is for the re-upload pipeline across the whole library.
- Deepgram (speech-to-text API): about $0.26 per hour — roughly $26 for 100 hours, plus your own engineering time to run every video through it and attach each file by hand.
- Rev AI captions: $0.25 per minute — about $1,500 for 100 hours.
- Rev human captions: from $1.99 per minute — about $11,940 for 100 hours, at 99% accuracy.
When transcription wins
Pay per minute when any of these apply:
- You need compliance-grade accuracy — legal, medical, or accessibility obligations where 'mostly right' isn't enough.
- The video is over 8 hours — Vimeo never auto-captions those, so re-uploading won't produce captions either.
- The audio fights the machine: loud music, heavy crosstalk, or degraded recordings.
- The spoken language isn't one of the ~90 Vimeo's Auto-CC supports.
- You need translated subtitle tracks, not just same-language captions.
When re-uploading wins
Everything else — which describes most back catalogues. Hundreds of videos, no caption files, clean spoken audio, and auto-caption quality that's good enough (and editable in Vimeo's transcript editor afterward). At library scale the per-minute route costs orders of magnitude more, and someone still has to attach every file.
What most libraries should do
Both, in order. Run the backlog through Recaption so every video gets auto captions for one flat fee, then send the exceptions — the showpiece course, the keynote, anything that failed on audio — to a transcription service. That's the cheapest path to a fully captioned library, and it puts the per-minute bill only where it's actually needed.
Questions.
- Are Vimeo's auto captions as good as Rev's human captions?
- No. Human captions claim 99% accuracy; auto captions are good on clean single-speaker audio and weaker with accents, crosstalk, or music. For most archives, auto captions plus a transcript-editor pass on the videos that matter is the right level.
- Deepgram is cheaper than Recaption. Why not use that?
- On raw transcription price, it is — if you can run the pipeline yourself: API keys, polling, then attaching one caption file per video by hand. Recaption is the no-engineering path: paste a token, review the split, get captioned copies. You're paying for not building that pipeline.
- Can I mix both routes in one library?
- Yes — they coexist naturally. Re-upload the bulk for auto captions, then order human transcription for the handful of videos that need it and attach those files to the same copies.