Free tool

Free AI Subtitle Generator

Most people watch with the sound off. Subtitles are not a nice extra any more - they are the difference between a video being watched and being scrolled past.

Drop in a video and the speech is recognised on your own machine. Edit any line you want, then burn the captions into the picture or download an SRT or VTT to upload alongside it.

Runs in your browser - your file is never uploaded

How it works

1

Choose your video

The audio is read on your machine. Nothing is uploaded anywhere.

2

Let it listen

Speech recognition runs in your browser and times every word.

3

Edit, then burn or export

Fix any line, pick a style, and burn in or download SRT/VTT.

More free tools

All of these run in your browser. No upload, no account, no watermark.
Video TrimmerCut without quality loss Video CompressorShrink for upload limits MP4 CompressorHit a target file size Video Collage MakerTwo to four clips in one Picture Video MakerPhotos into a slideshow Outro MakerAdd a branded end card Add Text to VideoBurn in a caption

FAQ

Is my video uploaded to transcribe it?

No. The speech recognition model runs inside your browser, so the audio never leaves your machine. The only thing that travels is the model itself, which is downloaded to you the first time - about 40 MB, then cached. That is a download in, never an upload out.

How long does it take?

Roughly one and a half times the length of the video on a typical laptop, so a one-minute clip takes around ninety seconds. It is slower than a cloud service because it is running on your own hardware rather than a datacentre GPU - that is the trade for it being free, private and unlimited.

Should I burn the subtitles in, or use the SRT file?

Burn them in when the video is going somewhere that strips caption tracks or gets re-shared - TikTok, Reels, Shorts, a message thread. Use the SRT when you are uploading to YouTube or a player that supports caption tracks, because viewers can then switch them off and search engines can read them.

Which languages does it handle?

English only at the moment. The model is an English-specific one, which is what keeps it small and fast enough to run in a browser tab.

Will it get every word right?

No automatic system does. It is accurate on clear speech and struggles with heavy background music, crosstalk and unusual names - which is exactly why every line is editable before you burn anything in.

Need to make the video, not just edit it?

Zappy generates video, images, voiceovers and music from a text prompt - 25 free credits every day, no software to install.

Open the studio