Video Transcriber
Upload a video or audio file and get the words back — a clean transcript, timestamped segments, and SRT or VTT captions. Nothing is stored.
Pick a file and press Transcribe.
You will get the full text, timestamped segments, and SRT or VTT captions.
Transcribing…
Longer recordings take longer — a ten-minute video is usually under a minute.
Work backwards from a video you already made
Every other AI tool here helps you build a video that does not exist yet. This one goes the other way. Point it at something you already published and you get the script back — which is usually the fastest way to make the next thing, because the hard part of a faceless video is not the edit, it is having something worth narrating.
Once you have the text, the rest of the site takes over. Feed it to Scene Splitter to get a shot list for a re-cut. Run the strongest ninety seconds through the Hook Generator to find the opening you should have used. Drop the timestamps into a description as chapters. Or take a long video's transcript and cut three Shorts out of the moments that already work.
What you actually get
A clean transcript, the same words split into timestamped segments, and caption files in SRT and VTT. Those two formats cover almost everything: YouTube takes either directly, and Premiere and Resolve both import SRT. Uploading real captions rather than letting a platform auto-generate them is one of the few small things that reliably helps — auto-captions are what get searched, and they are wrong often enough to matter.
About the 25 MB limit
That ceiling belongs to the transcription service, not to us, and it is measured in file size rather than length — so it depends entirely on what you upload. Spoken-word audio at a sensible bitrate fits roughly forty minutes into 25 MB. A 1080p video of the same length will not fit, because almost all of it is picture.
The fix is to export the audio on its own rather than to cut the video down. In almost any editor that is a single export preset, and the audio track of a video is typically a tenth of the file. A mono MP3 at 64 kbps is more than enough for transcription and will hold well over an hour.
Nothing is kept
The file goes straight through: sent when you press the button, transcribed, and never written down anywhere. It is not stored, not logged, and not used to train anything. What survives the request is a count — how many words came back, how many segments, how long it took — which is what tells us the tool is working without telling us what you said.
Common questions
Is my video stored anywhere?
No. The file is sent when you press Transcribe, transcribed, and then it is gone — it is never written to disk, never logged, and never used for training. Only counts survive the request: how many words came back and how long it took.
Why is there a 25 MB limit?
That is the ceiling on the transcription service itself, not a restriction we added. A 25 MB file is roughly 40 minutes of spoken-word audio at a normal bitrate. If your video is over it, export the audio track on its own — a 1080p video is almost entirely picture, and the audio inside it is usually a tenth of the size.
Can I get subtitles out of it?
Yes. Alongside the plain transcript you get timestamped segments, and you can download them as SRT or VTT — both formats YouTube, Premiere and DaVinci Resolve accept directly.
What languages does it handle?
It detects the language itself and transcribes in the language spoken, so you do not need to tell it anything. It does not translate — what was said is what you get back.
Does this use one of my generations?
Yes, one per transcription, from the same shared allowance every AI tool spends — free generations first, then any credits you have bought. A failed or timed-out transcription costs you nothing.