Transcribe audio to text
Transcribe any audio file to text in your browser: OpenAI's Whisper runs on your own computer, so the recording is never uploaded. Free, with timestamps, in 100 languages.
Drop a recording 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.
How it works
- Drop the recording (audio or video, up to 3 hours) onto the box above. Fast is free; Accurate makes fewer mistakes (Pro).
- Whisper transcribes it on your computer. The first time, the model downloads once; then it works offline too.
- Read along: click a line to hear it, and fix any word in place.
- Download the transcript as text, or as SRT or VTT subtitles with timestamps.
Transcription without the upload
Most online transcription services ask you to upload the recording to their servers, then wait. AudioRinse runs OpenAI's open Whisper model inside your browser instead: the audio is decoded and transcribed on your own processor and graphics chip, and the transcript is saved only on your device. Nothing is sent anywhere, so there's no upload to wait for and nothing for a service to keep.
That makes it fit recordings you shouldn't hand to a third party: meetings, calls, lectures, voice memos, interviews, a doctor's or lawyer's notes. After the first run it even works with Wi-Fi off.
Fast or Accurate
- Fast (Whisper small, 249 MB once) runs on the processor at about twice real time. It's free, as text or subtitles, with no limit.
- Accurate (Whisper large-v3-turbo, about 560 MB once) runs on the graphics chip and makes markedly fewer mistakes, especially with accents, names and background noise. Downloading its transcripts is Pro.
- Both write clean verbatim: most "um" and "uh" are left out, the words are kept as spoken, with punctuation and capitals.
Getting a good transcript
- Record close to the speaker. Quiet rooms beat expensive microphones.
- If the recording is noisy, run it through the background noise remover first: cleaner audio means fewer wrong words.
- Pick the language if you know it; detection reads the first 30 seconds.
- Check names and technical terms: Whisper spells them the way they sound. Click the line, listen, and fix it in place.
- Several people talking? The interview transcriber labels who says what.
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, with the Fast model: transcripts as text, SRT or VTT, with no limit and no account. Pro ($7.99 a month or $39.99 a year) adds the Accurate model, speaker labels, Word files and videos with burned-in subtitles. The same Pro covers every AudioRinse tool.
Which files can I transcribe?
MP3, M4A, WAV, FLAC, OGG and Opus audio, and videos (MP4, MOV, WebM, MKV): the sound is transcribed. iPhone voice memos work as they are.
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.
Does it work offline?
Yes. After the first run the page, the engine and the model stay in your browser: turn Wi-Fi off, reload the page and it works the same.
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
- Convert audio to textSpeech in any audio file, converted to text on your device.
- Convert MP3 to textAn MP3's speech as text, with timestamps.
- Transcribe a video to textWhat's said in a video, as a timestamped transcript.
- Convert video to textSpeech in any video, converted to text on your device.
- Add subtitles to a videoAuto subtitles, edited in place, burned in or as SRT.
- Auto captions for your videoCaptions for reels and videos, styled and burned in.
- Generate subtitles automaticallyTimed SRT and VTT subtitles from any video.
- Interview transcriptionSpeaker-labelled transcripts, made on your computer.
- Transcription for qualitative researchInterviews and focus groups, ready for NVivo and MAXQDA.
- Remove background noise from audioFans, traffic, hiss and room noise out of a voice recording.
- Remove background noise from videoClean a video's soundtrack. The picture is kept untouched.
- Voice isolatorKeep only the voice; everything else goes.
- Audio enhancerNoise removal + voice EQ + loudness, in one step.
- Voice enhancerClearer, fuller, evenly loud voice.
- Remove reverb from audioLess room, more voice: for recordings made in echoey rooms.
- Remove echo from audioFix the hollow, echoey sound of a big or empty room.
- Remove buzz and hum from audioElectrical buzz and hum out; the voice or music around it untouched.
- Remove silence from audioDead air out: every long pause cut to a set length.
- De-esserHarsh S sounds softened; the voice keeps its brightness.
- Fix clipped audioFlattened, crackly peaks from recording too loud, rebuilt.
- Mouth click removerLip smacks and tongue clicks out of a voice-over.
- Normalize audioPodcast, YouTube or broadcast loudness, true-peak safe.
- Remove vocals from a songThe vocals out of a song: a clean instrumental, and the acapella too.
- Split a song into stemsVocals, drums, bass and the rest, each on its own track.
- Remove music from a videoBackground music out of a video; the voices stay.
- Make a karaoke version of a songAny song as a karaoke track: the vocals out, the band left.
- Extract the vocals from a songThe vocals alone from any song: an acapella for remixes and covers.
- Make an instrumental of a songThe instrumental version of any song, the vocals taken out.
- Remove the drums from a songThe drums out of a song, for drummers to play along.
- Find the BPM of a songThe tempo of any song, with its key and Camelot code.
- Find the key of a songThe key of any song, with its Camelot code and tuning.
- BPM and key finderTempo, key and Camelot for a whole folder of songs, as a CSV.
- Tap tempoTap along on each beat and read the BPM.
- Camelot wheelWhich keys mix: pick one, or drop your songs onto the wheel.
- Audio to sheet musicA piano recording written out as a two-hand score.
- MP3 to sheet musicAn MP3 of piano music, written out as a printable score.
- Piano transcriptionPiano playing written out, both hands, in seconds.
- Video to MP3The sound of any video as an MP3. Free, no limit.
- MP4 to MP3Any MP4's sound as an MP3, in seconds.
- Video to audioAny video to MP3, M4A or WAV. Free, no limit.
- MOV to MP3iPhone and QuickTime videos to MP3.
- MP4 to WAVUncompressed WAV from any MP4, for editing.
- WebM to MP3WebM video or audio to MP3.
- MKV to MP3OBS recordings and MKV films to MP3.
- Extract audio from videoA video's sound track as a file, in its original quality.
- M4A to MP3Voice memos and iTunes songs as MP3.
- WAV to MP3WAV to MP3 at a fraction of the size.
- MP3 to WAVMP3 to uncompressed WAV, for editing.
- M4A to WAVVoice memos and M4A to uncompressed WAV.
- OGG to MP3Ogg Vorbis, Opus and voice messages to MP3.
- FLAC to MP3Lossless FLAC to MP3, tags and cover kept.