One button, then a list you can fix
Transcription is never perfect, so the result is not read-only. The caption list is editable in place: correct a name it misheard, split a line that ran long, and what you changed is what renders.
This is the difference between a transcript you have to take somewhere else to fix and captions you can actually finish in one sitting.
It runs locally, so the video never leaves
The speech engine and its model are fetched automatically the first time you use captions, about 85 MB, once. After that it works with no network at all.
That means footage under an agreement, an unreleased cut, or anything you would simply rather not hand to a third party can still be captioned. It also means captioning still works on a plane.
No per-minute billing to think about
Because the transcription happens on your own hardware, a long recording costs nothing beyond the time it takes. There is no meter running, so there is no reason to caption only the parts you think you will keep.
Caption first, cut second, and let the cues follow.
Cues follow your edit
This is where most tools quietly go wrong. If you trim a clip after transcribing, a line that spans the cut appears in both halves, and speech you trimmed away captions nowhere.
What you see in the preview is exactly what burns into the export, so there is no separate pass to re-sync captions after a change to the cut.
Captioning a walkthrough or a talking-head cut
Screen recordings and straight-to-camera pieces are the two cases where captions earn the most, because both are almost entirely speech and both are usually watched with the sound off.
Reel records the screen and edits the result in the same window, so the caption pass is one more button in a place you are already standing rather than a trip to another program.
Captions from the command line
The headless side of Reel captions a project with a single command, so a batch of recordings can be transcribed without anyone sitting in front of them.
Commands return one JSON object on request and fail loudly rather than quietly, which is what you want when the run is unattended.
Common questions
How do I add subtitles to a video without uploading it?
Use Reel's caption button. The speech is transcribed on your own machine, so the video is never uploaded and no account is involved.
Does offline transcription need an internet connection?
Only once. The speech engine and its model are fetched automatically the first time, about 85 MB, and after that captioning works with no network at all.
Can I correct a word the transcription got wrong?
Yes. The caption list is editable in place, and your correction is what burns into the export.
What happens to the captions if I trim the clip afterwards?
They follow the edit. A line spanning a cut appears in both halves, and speech you trimmed away captions nowhere, so there is no re-sync pass.
Is there a per-minute charge for long videos?
No. Transcription runs on your own hardware, so length costs time rather than money.
Can I generate captions from the command line?
Yes. The headless side of Reel captions a project with one command, and speaks JSON so a script can drive it.
Free and open source. Linux-first, with a Windows build too.