Transcribe video and audio on your device

Drop a recording and get a timestamped transcript and captions. Your video never leaves your computer: the speech model downloads once and runs in this tab.

Drop a video or audio fileor click to choose. MP4, MOV, WebM, MP3, M4A, WAV

Accuracy Language

Runs on your device. The speech model (OpenAI Whisper) downloads once, then works offline.

Caption style and frame

Style Frame Text size Height from bottom Pan (crop position) Words per caption Capitals Show app safe area

Safe-area shading is approximate: TikTok, Reels and Shorts move their buttons now and then.

Clip none

Click a line to start the clip there, then Shift-click (or long-press) the last line. Nudge the ends below.

In –

Out –

Exports at the frame and caption style above. No watermark.

Check and fix 0

Chars/line Lines Chars/sec

Shift timing

Two-point sync

For drift. Enter when an early line and a late line should appear (m:ss.s).

Line should be at Line should be at

YouTube chapters –

Breaks are suggested where the speaker pauses and a sentence ends. Rewrite the titles, then copy the list into your video description.

How many

Download

Project saves the transcript, edits and settings so you can pick up later. Your work is also kept in this browser automatically.

More free tools

YouTube transcriptCaptions YouTube already has, in seconds, searchable and downloadable.TikTok transcriptTikTok link to text, SRT or captions in seconds.Instagram transcriptReel link or saved Reel to text, SRT and captions.Facebook transcriptFacebook videos and Reels to text and subtitles.X (Twitter) transcriptVideos posted on X to text and subtitles.YouTube Shorts transcriptText of any Short, from its link, in seconds.Subtitle fixerOverlaps, flash cues, long lines, drift: found and fixed. SRT ⇄ VTT.Clip from transcriptClick sentences, export that clip as MP4 with its subtitles.YouTube chaptersChapter timestamps at natural pauses, checked against YouTube rules.Add captions to videoStyled, burned-in captions, 9:16 crop and safe-area guides.Loudness checkerLUFS and true peak vs YouTube, podcast and broadcast targets.Audio syncLine up a separate recorder with your camera automatically.Compress videoFit a video under 10 MB, 25 MB or any size, no upload.Extract audioVideo to M4A (lossless copy) or WAV, instantly.Thumbnail frame pickerSharpest frames from a video, previewed at YouTube sizes.Silence removerFind dead air; export an EDL timeline or a tightened WAV.

Questions

How accurate is it?

It uses OpenAI's open Whisper model. Clear speech in a quiet room comes out very close; heavy accents, crosstalk and music lower it. Every word is editable and the timing follows your edits.

How long does it take?

With a recent graphics chip, usually faster than the recording plays. On older laptops it can take longer than the recording. The first run also downloads the model, once.

Which languages?

Whisper understands about 100 languages and detects the language by itself. English is the most accurate.

Is there a length limit?

No fixed limit, but your browser has to hold the audio in memory. Recordings up to about an hour work on most laptops.