Generate subtitles automatically

Generate subtitles for any video automatically: Whisper writes timed SRT and VTT files in your browser, in 100 languages. Free, editable, nothing uploaded.

AudioRinse · subtitle generatorOn-device · 0 B uploaded

Drop a video here

MP3 · WAV · M4A · MP4 · MOV · WEBM — up to 3 hours. Nothing is uploaded.

The model downloads once and stays in this browser, so later transcripts start at once and work with Wi-Fi off. Your recording is never uploaded. Free: Fast transcripts without speaker labels, as text, SRT or VTT, no limit. Pro: Accurate, speaker labels, Word, NVivo and MAXQDA files, and the video with the subtitles burned in.

How it works

  1. Drop the video (or its audio) onto the box above. Fast is free; Accurate makes fewer mistakes (Pro).
  2. Whisper writes the subtitles on your computer, cut into readable captions.
  3. Check them against the video and fix any word.
  4. Download SRT or VTT, or the video with the subtitles burned in (Pro).

SRT and VTT, ready for any player

The subtitles come out as standard SRT and WebVTT files: numbered (SRT) cues, each with a start and end time to the millisecond and at most two lines of about 42 characters. YouTube, Vimeo, VLC, Premiere, DaVinci Resolve, Final Cut and CapCut load them as they are.

How the captions are cut

  • At most two lines of about 42 characters, the length broadcasters' guidelines settle on, and no more than 7 seconds each.
  • A new caption after a sentence ends, or after a comma once a caption is mostly full, so a line rarely stops mid-phrase.
  • Two lines are balanced: 21 and 21 characters rather than 38 and 4.
  • Each caption stays up at least a second where the next one leaves room.

Subtitles in another language

Whisper writes subtitles in the language spoken, in about 100 languages. To subtitle a video in another language, translate the SRT afterwards: most translation tools keep its timing lines intact.

Questions

Is my recording uploaded anywhere?

No. OpenAI's open Whisper model runs inside your browser, on your own computer. The recording and the transcript never leave it; the only download is the model, the first time. The data-flow statement lists every request the page makes.

Is it free?

Yes: SRT, VTT and text with the Fast model, no limit, no account. Pro ($7.99 a month or $39.99 a year) adds the Accurate model, which makes fewer mistakes, and the video with the subtitles burned in.

SRT or VTT?

SRT for YouTube, video editors and most players. VTT for web players (HTML5 video, course platforms). They hold the same captions.

Can it generate subtitles from an audio file?

Yes. Drop the audio (a podcast, a recorded talk) and download the SRT or VTT; only burning into a picture needs a video.

How accurate is it?

Accurate (Whisper large-v3-turbo) got 8.3% of words wrong on three recorded hour-long interviews with professional reference transcripts, on a par with desktop Whisper tools on the same files. Fast (Whisper small) makes more mistakes, mostly on names, accents and noisy recordings. Clear speech close to the microphone helps both.

How long does it take?

Fast runs on your processor at about twice real time: a 30-minute recording takes about 15 minutes on a recent laptop. Accurate runs on the graphics chip through WebGPU, about as long as the recording on a 2020 laptop's built-in graphics and faster on newer ones. The first time also downloads the model.

Which languages does it understand?

Whisper transcribes around 100 languages, most accurately English and the major European languages. Pick the language in the menu, or let it detect it from the first 30 seconds.

How long can the recording be?

Up to 3 hours, audio or video. On a computer with little memory the page takes less, and says so.

Other tools