Free caption generator: auto subtitles for any video
Drop a video and get captions in a minute. Whisper writes them right here in your browser, splits them into short lines you can read at a glance, and shows them on the video so you can fix any word. Download an SRT file or burn the captions into the video. Free, no watermark, nothing uploaded.
Your file is read by this browser tab and never sent anywhere. The first run downloads the AI model from Hugging Face and caches it, so later runs start faster and work offline.
How to generate captions for a video
- Drop the video into the box. It starts playing in the preview while Whisper listens to the sound track.
- Choose Short captions for vertical social video or Standard for YouTube and subtitles. You can switch after the captions are made.
- Read through the list. Click a timestamp to play that line, type over any word that is wrong, and delete lines you do not want. The preview updates as you type.
- Download the SRT or VTT file, or pick a style and click Burn into video to save a copy with the captions drawn on the picture.
SRT generator or burned-in captions?
An SRT file is a small text file of timed captions that sits next to the video. Viewers can turn
it on or off, YouTube and Vimeo index it, and every editor imports it. Upload it with the video on YouTube
(Subtitles, then Upload file) or drop it onto a timeline in Premiere Pro, Final Cut Pro or DaVinci Resolve. WebVTT
is the same thing for web players and the HTML5 <track> element.
Burned-in captions, sometimes called open captions or hardcoded subtitles, are part of the picture. They show up everywhere, which is why short-form video uses them: most people scroll with the sound off, and TikTok, Reels and Shorts do not all read caption files. The trade-off is that they cannot be edited or switched off afterwards, so keep the SRT as well.
How the captions are made
The page decodes the sound with your browser's own media decoder and runs OpenAI Whisper through Transformers.js, on the graphics card through WebGPU when the browser has it and on the CPU otherwise. Whisper returns sentence-length segments with start and end times. Those are too long to read on screen, so each one is split into captions that fit the chosen length, preferring to break at the end of a sentence, and each caption gets its share of the segment's time.
Burning in is done the same local way: the browser plays the video once, draws every frame and its caption onto a canvas, and records that with the original sound. No ffmpeg download and no server, so a two-minute clip takes about two minutes and stays on your computer. Keep the tab in front while it records.
Tips for accurate captions
- Pick the language yourself for short clips or clips that open with music; automatic detection listens to the first seconds only.
- Switch to Whisper Small for noisy rooms, strong accents or lots of names. It is a larger download but makes fewer mistakes.
- Names and brand words are where Whisper slips most. Search the list for them before you export.
- Keep captions short for vertical video. One line in the lower third leaves the face and the app's own buttons clear.
When the desktop app is the better tool
This page captions a video you already have. If you are cutting that video from a pile of footage, the slow part comes earlier: finding the clips. The Reelary desktop app indexes your whole footage library on your own machine, so you can search it by what is in the picture ("close-up of hands on a keyboard") or by what someone said, and drop the results straight onto a timeline.
Its editor adds a subtitle track to the cut and exports an MP4 with the captions burned in, or hands the timeline to Premiere Pro, Final Cut Pro or DaVinci Resolve as an FCPXML or XML project. See how local AI video search works, or just get the plain transcript with the video to text tool.
The desktop app captions the cuts you assemble from your library and exports them with subtitles burned in or as an editable timeline. Reelary is free to download for Mac and Windows; indexing and search have no file limit.
Questions
Is this caption generator free?
Yes. There is no sign-up, no watermark on burned-in video and no per-minute charge, because the speech recognition runs on your own computer. Files up to 1 GB and 30 minutes are accepted, and burning captions into the video works up to 10 minutes.
Is my video uploaded?
No. The browser reads the sound track, the Whisper model runs inside this tab, and the captioned video is recorded by the browser too. The only download is the model itself, fetched once from Hugging Face and cached.
How do I make an SRT file from a video?
Drop the video into the box above and wait for the captions to appear. Fix any words Whisper got wrong directly in the list, then click SRT. The file works in YouTube, Vimeo, Premiere Pro, Final Cut Pro, DaVinci Resolve and most players.
Can I burn the captions into the video?
Yes. Pick a style, size and position, then click "Burn into video". The browser plays the video once and records it with the captions drawn on every frame, so it takes as long as the video. Chrome, Edge and Safari save an MP4; Firefox saves a WebM.
What is the difference between short and standard captions?
Short captions are one line of up to 32 characters, the fast-moving style used on TikTok, Instagram Reels and YouTube Shorts. Standard captions allow two lines of up to 42 characters, which is the usual guideline for YouTube, broadcast and accessibility subtitles.
Which languages can it caption?
Whisper recognises 99 languages and detects the language on its own. The captions are written in the spoken language; this tool does not translate.
How is this different from the video to text tool?
Both use the same Whisper model. Video to text gives you a transcript to read or paste. The caption generator breaks the speech into short timed captions, shows them over the video so you can correct them, and can burn them into a new copy of the video.
More free tools
Published October 10, 2026 · Models we use