Most transcription services need your file on their servers first. Burncaps runs the speech model inside your own browser, so a confidential recording never leaves your machine (a faster, more accurate speech service is offered on phones and for some languages; it gets only the sound, and choosing your device, or ticking Keep everything on this device, means nothing is sent). Transcripts and subtitle files are free without limit, and you can verify the claim yourself in about ten seconds.
Transcribe privately, freeLive preview, rendered on this page by the same engine that burns your video.
Press F12 in Chrome or Edge, choose Network, then load your file into the page. Everything a website sends or receives shows up in that list.
You will see the Whisper model download once from Hugging Face (about 190 MB with a graphics chip, 80 MB without), and then nothing. No upload of your file appears, because none happens (had you chosen our speech service, you would see pieces of sound going out, never the picture).
Once the model has cached, disconnect from the internet and transcribe. It still works, which is only possible if the processing is happening on your own hardware.
If you record client calls, medical or legal discussions, unreleased material, or anything under an NDA, uploading it to a transcription service is a disclosure to a third party, whatever their privacy policy says. Processing that never leaves your device removes the question entirely. There is no server of ours that could be subpoenaed, breached or used for training, because there is no server.
| What matters | Burncaps | Cloud transcription service |
|---|---|---|
| Where your file goes | Nowhere. It stays on your device | Uploaded and stored on their servers |
| Used to train their models | Impossible, we never receive it | Sometimes, depending on their terms |
| Works offline | Yes, once the model is cached | No |
| Length limits | None | Minutes-per-month caps |
| Price | Transcripts and subtitle files free, unlimited. Burned-in video: free, with no mark on your first two, then a small mark or $49 once | Per minute or monthly subscription |
Yes. The model file downloads into your browser once and then executes there using your own processor. You can confirm it by transcribing with your internet disconnected after the first run.
It uses the same Whisper model family that many cloud services are built on. The standard model is about 190 MB on a computer with a graphics chip (80 MB without); the large-turbo model for harder audio is included too, both running locally.
No. Because there is no server bill to protect, an hour-long recording is fine. Longer files simply take longer to process on your machine.
SRT, VTT, ASS and plain text, or captions burned directly into a new MP4. Everything is generated and saved on your device.
A cookieless page-view counter that cannot identify anyone, and nothing else. No accounts exist, so there is no personal data to hold.
What it is: a browser tool that burns animated word-by-word captions into a talking video, or cuts it to its best 30 or 60 seconds.
Price: free with every feature, for as many videos as you like. The first two carry no mark; after that each carries a small burncaps.app credit, which Pro removes for $49 once. No subscription.
Privacy: the video never leaves your browser. The sound leaves only if you choose our speech service.
Files: MP4 or MOV up to ten minutes in; an MP4 with the captions burned in out. SRT, VTT, ASS and text files are free without limit.
Languages and styles: captions in the language spoken, twelve Indian languages heard on your device, translation into 37 languages, 20 caption styles.
Made by: Jordan Labs Ltd, UK company no. 17358646.
Against other tools: twelve caption tools compared, from the vendors' own pages.