Skip to content

Auto caption generator

Turn what's said in a video into timed captions, for free: edit them, download SRT or VTT, or burn them into a vertical video. It runs in your browser, so your video is never uploaded.

MP4, MOV, WebM, MP3 or WAV, or drop one on the player. It's transcribed on your device: nothing is uploaded.

An example: 8 seconds of IMPAULSIVE

Settings

Free, with nothing to sign up for. The speech model is a one-time download of about 76 MB; after that it's kept in your browser.

How to generate captions for a video

  1. Choose your video or audio file, or drop it on the player. Pick a part if you don't need all of it.
  2. Pick the language and how long each caption should be: a few words at a time for TikTok, Reels and Shorts, or full subtitle lines for YouTube.
  3. Generate the captions, play the video to see them, and fix any word that's wrong.
  4. Download an SRT or VTT file, or burn them into a vertical video in the 5 caption styles clips use.

A few words at a time, or full subtitle lines?

The two caption lengths and what each is for
CaptionsEach oneFor
A few words at a timeUp to 4 words, cut at commas and pausesTikTok, Reels and Shorts, burned into the video
Full subtitle linesUp to 2 lines of 42 characters, a sentence at a timeYouTube videos, courses and interviews, as a subtitle file

Short videos are watched on a phone, often with the sound off, so their captions are big, a few words at a time, and on the picture. For more on styling them, read how to caption short videos; for the files themselves, what an SRT file is.

How accurate auto captions are

The captions come from Whisper, OpenAI's speech recognition model, in a version small enough to run in a browser. Clear speech with little background noise comes out close to word for word. Names, brands, numbers and people talking over each other are where it slips, so read the captions through before you post: each line can be fixed in place. Convert an SRT to VTT, or shift its timing, with the SRT to VTT converter. The same speech model powers the bleep censor, which finds the swear words and bleeps them before you caption the clip.

Questions

How do I add captions to a video automatically?

Choose the video, pick the language (or let it detect it) and press Generate captions. Read them through, fix any word, then download an SRT or VTT file, or press Burn them into a vertical video to put them on the picture.

Is my video uploaded anywhere?

No. The captions are made by a speech recognition model (OpenAI's Whisper) running in your browser. The model is downloaded once, about 76 MB, and kept in your browser; your video never leaves your device.

Which languages does it understand?

About 100, the languages Whisper was trained on. It can detect the language for you, but picking it is more reliable for short clips or when people switch languages. English, Spanish, Portuguese, French and German come out best.

Can I edit the captions?

Yes. Every line can be changed before you download it, and its timing stays where it was. Names, brands and numbers are worth a check.

How long a video can it caption?

Up to 30 minutes at a time, from a video of any length: pick the part. On a computer's processor it takes about half as long as the part plays, and much less where the browser can use the graphics card.

How do I put the captions on TikTok, Reels or YouTube?

For short videos, burn them into the picture, so they look the same everywhere and can't be turned off. For a YouTube video, you can also upload the SRT file in YouTube Studio's Subtitles section, as a track viewers can turn on.

Captions on every clip, automatically

Paste a YouTube link and Clyper finds the best moments, frames them vertically and captions every word in the style you pick: up to 30 clips per video.

or upload a video

Guides

More free tools