Video to WAV or M4A, in your browser. Pull a clean audio track out of any video file, ready for a transcription service.
WAV is uncompressed — what transcription services want.
MP3 and FLAC are off because no browser can encode them without a bundled codec, and shipping one to everybody for two formats isn't a good trade.
SAMPLE RATEMatches the source
CHANNELSCLEAN-UPDecoding happens on this device. The file never leaves it.
MP4, WebM, MOV and most audio files. The track is decoded straight from the file on your disk — nothing is uploaded.
WAV for transcription and editing, M4A when size matters. Normalise the loudness, strip long silences, or filter out rumble.
Unless you strip silence, the timing matches the video exactly, so anything you build on the audio lines up when you go back.
Something not covered? [email protected] reaches a person.
Yes — there is no stream copy here. The browser decodes the track to raw samples and this re-encodes them, to PCM for WAV or to AAC for M4A. WAV from a compressed source can't add back what compression removed, but it won't lose anything more. If you need a bit-for-bit copy of the original track, that needs a demuxer, which means a desktop tool.
Because most transcription services and every podcast host expect a consistent loudness, and −16 LUFS is the usual target for stereo. It's measured properly, to ITU-R BS.1770 with gating, so long pauses don't drag the reading down and make the result too loud. If reaching the target would clip, the gain is held back to leave a decibel of headroom and the result says so.
Yes, and that's the point of the warning next to it. Removing four seconds of silence moves everything after it four seconds earlier, so a transcript made from the stripped file will not line up with the original video. Leave it off if you're going to sync anything back.
About twenty minutes. The whole track is decoded into memory as raw floating-point samples — twenty minutes of 48 kHz stereo is already around 460MB — so past that the tab runs out of room. It refuses with a message rather than crashing. There's no upload, so file size on disk matters much less than duration.
No browser can encode either one. WAV we write ourselves — it's a 44-byte header around raw samples — and AAC is available because the browser's own recorder supports it. MP3 and FLAC would each mean shipping a codec to every visitor, which is a poor trade for two formats when WAV already does the job.
Nudge an entire subtitle file forward or back when your export drifted out of sync.
Shift timingsWord count, character count, reading time, and words per minute for any transcript.
Count wordsDrop in a screenshot and see what TikTok, Reels, and Shorts will cover up.
Check placementYou can skip a step. Upload the video here and we'll transcribe it, time it to the word, and burn the captions in — five dollars a month.