How to add captions to a video
Most people scroll through short videos with the sound off, so captions decide whether they stop or keep going. This tool listens to your video, writes the captions word by word, lets you fix anything it got wrong, and burns them into the picture in the style you choose. It all happens in your browser: your video is never uploaded.
- 1
Add your video
Drop in an MP4, MOV or WebM file up to 10 minutes long, or a sound file such as a podcast clip. The first time, the tool asks before loading its speech model (about 110 to 145 MB, depending on your browser). After that it starts straight away.
- 2
Let it write the captions
The tool finds where people are speaking, then turns the speech into words with timings for every single word. A one-minute clip usually takes a few seconds on a recent laptop.
- 3
Fix words and pick a style
Click any word to correct it, use Find & replace for names and brands, and split or join captions. Then choose a style, a font, how many words appear at once, the size, position and colors. The preview shows exactly what you will get.
- 4
Export
Download the video with the captions burned in, or download an SRT, VTT or plain-text file to upload alongside your video instead.
Six caption styles, one click each
Each style is tuned for a different kind of video. You can change the font, colors and size of any of them.
- Bold pop: big, outlined, all-caps words where the word being spoken pops in yellow. The look most short-form creators use.
- Karaoke: words light up as they are spoken, so viewers can follow along. Good for tutorials and storytelling.
- Classic: plain subtitles on a dark band, for interviews, documentaries and anything that should feel calm.
- Boxed: white words on a colored box, in your brand color if you like.
- Outline: tall, condensed text with a clean outline and no highlight.
- Minimal: small text with a soft shadow that stays out of the way.
Set words per caption to 1 for the fast, one-word-at-a-time look, or up to 6 for calmer subtitles. Turn on the safe-area guide to keep your captions clear of the buttons and descriptions that TikTok, Reels and Shorts lay over the bottom and right of the screen.
How accurate are the captions?
The captions are made by Whisper, a speech-recognition model from OpenAI, running on your own device. On clear speech it is very accurate: in our tests on recorded talking clips it got every word right and placed captions within a fraction of a second of the speech. To make the timing tighter, the tool first finds exactly where the voice starts and stops, then trims each caption to match.
It is still not perfect, and no automatic captions are. Names, brand names, slang, heavy accents, music under the voice and people talking over each other all make mistakes more likely. That is why every word can be edited before you export, and why we suggest reading your captions once before posting. Captions are English only for now.
Burned-in captions or a subtitle file?
Burned-in captions are part of the picture. They look the same everywhere, keep your styling, and need nothing from the platform, which is why creators use them on TikTok, Reels and Shorts. The trade-off is that viewers can’t turn them off and search engines can’t read them.
A subtitle file (SRT or VTT) is uploaded next to the video, and viewers switch it on and off. YouTube, Vimeo and most video editors accept them, and they are better for long videos and accessibility. Our guide to SRT and VTT files explains the difference. If you only need the file, the SRT Generator puts the downloads first.
Private by design
Most caption apps upload your video to a server, transcribe it there, and keep a copy for a while. Here the speech model comes to your browser instead, once, and your video stays on your device the whole time. Your edits are saved in your own browser so a reload doesn’t lose them, and you can delete the saved model any time on the Saved AI models page. How it works explains how to check this yourself.
After exporting, you can keep going: compress the captioned video for sharing, or trim the start and end.
Frequently asked questions
Is it really free?
Yes. Captions, styles and all three subtitle formats are free, with no watermark and no sign-up.
Does my video get uploaded?
No. The speech model is loaded into your browser once, and your video is processed on your own device. It is never sent to us or anyone else.
Which languages are supported?
English only for now. The model is trained for English speech, and other languages would give poor results, so we do not offer them yet.
Why is there a download the first time?
The captions are made by an AI model that runs on your device rather than on a server. Your browser downloads it once (about 110 to 145 MB depending on the browser) and keeps it, so next time it starts in a second or two.
How long can my video be?
Up to 10 minutes. On a recent computer with Chrome, Edge or Safari, captions for a minute of speech take a few seconds. Browsers without fast graphics support, including Firefox, use a slower mode that takes roughly a third of the video’s length.
Can I edit the captions after they are made?
Yes. Click any word to change or delete it, use Find & replace for repeated names, split or join captions, and move a caption 0.1 seconds earlier or later. Your edits are kept in your browser if you reload the page.
Will the captions look the same on TikTok and Instagram?
Yes, if you burn them in: they become part of the video, so every app shows them exactly as you styled them. Turn on the safe-area guide to keep them clear of each app’s buttons.