Is the video uploaded?

No. The speech recognition, the drawing and the encoding all run in your browser. The only download is the speech model itself, from huggingface.co, the first time; your browser keeps it for next time.

What is the difference between burned-in captions and switchable subtitles?

Burned-in captions are drawn into the picture, so they show everywhere - on TikTok, Instagram, YouTube Shorts, in any player - and cannot be turned off; the video is re-encoded to do it. Switchable subtitles are stored as a separate track in an MKV file, which VLC, Windows Media Player, mpv, Plex and Kodi let viewers turn on and off; nothing is re-encoded, so it is quick and loses no quality, but social apps and most phones ignore the track.

How accurate are the captions?

Whisper is very good with clear speech, and the Accurate model does better with accents, music or several people talking, at the cost of a bigger download and slower work. Names and unusual words are the usual mistakes, which is why every caption can be edited before saving.

How long does saving take?

Burning captions in re-encodes every frame, usually faster than the video plays on a computer with hardware video encoding (most do). Switchable subtitles take a few seconds, since nothing is re-encoded.

Stays on your computer To make captions from the speech, downloads a Whisper model from huggingface.co the first time: about 40, 80 or 250 MB depending on the model (80 MB by default)How this works

Free tool to add subtitles to a video

Make captions from what is said in a video, edit them, style them, and burn them into the picture - in your browser, so the video never leaves your computer

Drag & drop a video here, or browse

MP4, MOV, WebM · TikTok, Reels, Shorts, lectures, interviews · drop an .srt or .vtt with it to use those captions · nothing is uploaded

About adding subtitles to a video

Most videos are watched with the sound off, so captions burned into the picture are what make a clip on TikTok, Instagram Reels or YouTube Shorts readable at all. The online tools that make them usually upload the video, put a watermark on the result, or stop after a minute or two unless you pay.

This page listens to the video with Whisper, OpenAI's open speech model, running on your own computer, and turns what is said into captions with a time for every word. Check and correct them in the list, choose a style - big words a few at a time with the spoken word lit up, as on TikTok, or classic subtitles along the bottom - and watch the preview over the video. Saving draws the captions onto every frame and writes a new MP4 with your browser's own video encoder, with no watermark and no length limit beyond what your computer can hold. A subtitle file you already have can be used in place of the speech, and the captions can also be saved as a .srt or .vtt, or added as a subtitle track that players switch on and off. It is free to use, with no account and no upload limit, and it runs entirely in your browser: nothing is uploaded, and while the page stays open, a feature you have already used keeps working offline.

What you can do

Make captions automatically from the speech in a video, in about 100 languages
Translate the speech into English captions
Edit each caption's words and times, add and delete captions
Social style: a few big words at a time, with the word being spoken highlighted
Classic style: one or two lines along the bottom, on a dark box or with an outline
Choose the font, size, position and colours, with a live preview over the video
Burn the captions into the picture and save an MP4 with no watermark
Add the captions as a switchable subtitle track in an MKV, without re-encoding
Use an existing .srt or .vtt file, and save the captions as .srt or .vtt

How to add subtitles to a video

Drop the video onto this page.
Press Make captions from the speech (or load a subtitle file).
Fix any words the captions got wrong, and adjust the times if needed.
Pick a style and adjust it while watching the preview.
Press Save video with captions.

Questions

Is the video uploaded?: No. The speech recognition, the drawing and the encoding all run in your browser. The only download is the speech model itself, from huggingface.co, the first time; your browser keeps it for next time.
What is the difference between burned-in captions and switchable subtitles?: Burned-in captions are drawn into the picture, so they show everywhere - on TikTok, Instagram, YouTube Shorts, in any player - and cannot be turned off; the video is re-encoded to do it. Switchable subtitles are stored as a separate track in an MKV file, which VLC, Windows Media Player, mpv, Plex and Kodi let viewers turn on and off; nothing is re-encoded, so it is quick and loses no quality, but social apps and most phones ignore the track.
How accurate are the captions?: Whisper is very good with clear speech, and the Accurate model does better with accents, music or several people talking, at the cost of a bigger download and slower work. Names and unusual words are the usual mistakes, which is why every caption can be edited before saving.
How long does saving take?: Burning captions in re-encodes every frame, usually faster than the video plays on a computer with hardware video encoding (most do). Switchable subtitles take a few seconds, since nothing is re-encoded.

Built by John McMillan, a senior full-stack engineer. Open to remote roles: hire me.

Report a problem or suggest a feature