Xargaar ToolsCaptions for your video, made on your device.
Drop a video or audio file. The page listens to it, writes the captions, and lets you fix any word before you download SRT or VTT. Your file is never uploaded, and there is no account.
Three steps, no account
Any video or audio your browser can play. It is read on your device and never sent anywhere.
The speech model downloads once (the size is shown first) and is kept in your browser. English first. Other languages are in beta.
Edit the text and timing next to the video, then download SRT or VTT for YouTube and most other video platforms, editors and players.
Private by design
Interviews, client calls and unreleased videos often should not sit on someone else's server. After the model is downloaded, this page needs no connection to make captions. The model runs inside your browser and the page has no upload step.
What to expect
The free version captions the first 3 minutes of each file with the Tiny model. It is fast but makes more mistakes than larger models, so clear audio works best. Speed depends on your computer; a few minutes of audio can take about as long as it lasts on a slow laptop.
No length limit. Works offline.
- Captions for files of any length, not just the first 3 minutes.
- Tiny models included for English and multilingual beta.
- Burn captions into a video file (WebM) on your device.
- All models included, so it works with no internet after you unzip it.
- One-time price, no subscription.
Straight answers
Is my video uploaded?
No. The file is decoded and transcribed in your browser. The only download is the speech model, from this same site.
How big is the first download?
The size is shown above the file box before you start. The Tiny model is about 41 MB. It is stored in your browser, so the next time it starts straight away.
How accurate are the captions?
Good on clear speech, weaker with background noise, heavy accents, music or several people talking over each other. Always review before you publish. Other languages than English are beta and less accurate.
Which formats can I download?
SRT, WebVTT and plain text. The offline pack can also burn the captions into a WebM video.
Why is there a free limit?
Running the model uses your computer, not ours, so the limit is not about cost. The free version shows you the quality on the first 3 minutes. The pack removes the limit and includes local launch files.
Which browsers work?
Current Chrome, Edge, Firefox and Safari on a computer. Phones can work for short files but may run out of memory on long ones.
What are the models and licenses?
Speech recognition uses OpenAI Whisper models (Apache 2.0 on their model cards, MIT in the original code) converted to ONNX by Xenova, run with Transformers.js (Apache 2.0) and ONNX Runtime Web (MIT). See NOTICES.txt.