WAV transcription

WAV to text, with the file sizes explained

Convert WAV to text straight from the .wav file, with no format conversion first. WAV is uncompressed, which makes it the one audio format where size, not length, is usually what stops you — so this page tells you where that line falls before you start the upload.

Lossless in, plain text out About 99 minutes per WAV TXT / SRT / VTT

Drop your WAV file here

WAV, MP3, M4A, FLAC, OGG, WebM

Language detected automatically
TXT · SRT · VTT
TimestampsSpeaker turns
Transcribe my WAV
Why WAV

Where .wav files come from, and why the format survives

WAV is what you get when a device decides not to make a decision for you. Windows Sound Recorder, Audacity, handheld field recorders, conference room hardware, and almost every professional interface write WAV by default, because it stores exactly what the microphone heard with nothing thrown away.

The format dates to 1991 and has outlived nearly everything released alongside it. It has no patents, no licensing, and no variants to get wrong — every piece of audio software ever written can open a .wav file. That reliability is why studios and archives still standardise on it.

It also means a WAV file is enormous. Uncompressed CD-quality audio runs about 10 MB per minute, roughly ten times an equivalent MP3, and that single fact drives most of what follows on this page.

Nothing was discarded

WAV stores raw PCM samples. No psychoacoustic model decided which parts of the sound you would not miss, because no compression happened at all.

Ten times the size

About 10 MB per minute at CD quality. An hour-long interview lands near 620 MB — a size that matters when you go to upload it.

Universally readable

No codec to install, no licensing, no variants. This is why recorders and archival systems still default to it three decades on.

Size limits before time limits

For every other audio format the length cap binds first. For WAV it is the file size — which is what the table below is for.

Definition

What converting WAV to text involves

Converting WAV to text means running speech recognition directly on the file's uncompressed PCM audio. Nothing is converted beforehand: the recogniser reads the samples as they are, and returns the spoken words as timestamped text.

Four things about converting WAV to text are worth knowing before the first upload, and one of them contradicts what most people assume.

  • No format conversion needed You do not need to turn a .wav into an MP3 first. Converting to a lossy format before transcription can only remove information the recogniser might have used.
  • Lossless helps less than you would think The gap between a 128 kbps MP3 and an uncompressed WAV of the same recording is small. Microphone distance and background noise move accuracy far more than the file format ever does.
  • Recognition works at 16 kHz anyway Speech models operate on roughly 16 kHz mono internally. A 96 kHz 24-bit master is downsampled before anything is recognised, so the extra data only slows your upload.
  • Timestamps come from the audio Each line is stamped with when it was spoken. That is what lets a WAV transcript be exported as SRT, useful later if the audio is married to video.
How it works

Transcribe WAV to text in four steps

Everything runs in the browser. Nothing to install, and no conversion step before you begin.

1

Upload the .wav as it is

Drag the file straight in. It uploads directly to storage from your browser, which matters more here than for other formats — WAV files are large enough that an intermediate size cap would stop most of them.

2

The audio is read directly

PCM samples go to the recogniser without an intermediate encode. Nothing is stripped, degraded or re-compressed on the way in.

3

Speech recognition runs

The spoken language is detected across 100+ languages, and the transcript comes back split into timestamped segments you can read and correct.

4

Export what you need

Plain text for documents and notes, or SRT and VTT if the recording will later be paired with video.

A ten-minute WAV usually comes back in under a minute. The upload itself takes longer than the transcription for most files, simply because there are more megabytes to move.

Worth knowingIf your WAV is close to the size ceiling, convert it to FLAC before uploading rather than to MP3. FLAC is lossless too — it just stores the same samples more efficiently, at roughly half the size.
Worth knowingRecording something new specifically to be transcribed? 16 kHz mono is all the recogniser uses. It sounds thin to the ear but transcribes identically, at a fraction of the file size.
Where WAV shows up

The recordings that arrive as .wav

These share a pattern: the device chose the format, and it chose fidelity over convenience.

Handheld field recorders

Journalists and researchers record to WAV because it is the format the device trusts. A one-hour interview arrives around 600 MB.

Audacity projects

Exported as WAV by default. Often already cleaned up, which makes them among the better files to transcribe.

Conference room systems

Installed audio hardware writes uncompressed. The sound is usually excellent — a ceiling microphone still beats a phone across the table.

Legal and medical dictation

Uncompressed is often required by policy, because compression is a form of alteration. Accuracy expectations here are also the highest.

Podcast masters

The edit is kept as WAV and the MP3 is only made for distribution. Transcribe the master and you skip a generation of loss.

Archival and oral history

Collections standardise on WAV for preservation. Text is what finally makes decades of tape searchable.

30-minute interview About 310 MB at CD quality. Uploads and transcribes without trouble; costs 15,000 credits.

90-minute lecture About 930 MB — close enough to the 1 GB ceiling to be worth converting to FLAC first.

2-hour recording About 1.2 GB, which is over the limit. Convert to FLAC or mono and the same audio fits comfortably.

5-minute voice memo About 52 MB. Well inside the free tier, which covers roughly nine and a half minutes of uncompressed audio.

Input and output

What goes in and what comes out

WAV and other audio

MP3, M4A, FLAC, OGG and WebM all work the same way. FLAC is the useful one to know about — lossless like WAV, at about half the size.

TXT for reading

Plain text with or without timecodes. This is what most WAV transcripts become: notes, quotes, searchable records.

SRT and VTT

Timed files, useful even for audio-only recordings if the material will later be paired with video or published with a player.

100+ languages

Detected automatically. A recording that switches language partway through is transcribed in each language as spoken.

Nothing is written back into your WAV file — the transcript is a separate document, and your original recording is left untouched.

Accuracy

What actually determines a WAV transcript's quality

There is a widespread assumption that uncompressed audio transcribes noticeably better, and it shapes how people approach WAV to text. It does not, and knowing why saves you from optimising the wrong thing.

  • Lossless is not the lever A 128 kbps MP3 and an uncompressed WAV of the same recording transcribe almost identically. Compression discards frequencies above what speech occupies; the recogniser was never using them.
  • Microphone distance is A recorder two metres away captures the room along with the voice, and no amount of bit depth recovers what the room smeared. This single factor outweighs format entirely.
  • Sample rate barely matters Recognition runs at roughly 16 kHz. Anything above that is discarded before the model sees it, so 96 kHz gains nothing over 44.1 kHz.
  • Clipping cannot be undone If the level was set too hot and the waveform is squared off at the peaks, those words are gone. Uncompressed only means the distortion was preserved faithfully.
  • Background music still hurts Music under speech confuses recognition more than steady noise. Strip it before transcribing — lossless or not makes no difference here.
File size

How long a WAV can be before the size limit stops it

This is the one place WAV to text behaves differently from every other audio format. Paid plans allow 120 minutes or 1 GB per file — and because uncompressed audio is so large, WAV is the only format where the gigabyte runs out first. Roughly 99 minutes, not 120. Figures are for stereo unless noted.

Format and qualityBitratePer minute1 GB holdsWhat to do
WAV 44.1 kHz 16-bit (CD)1,411 kbps10.3 MB~99 minutesFine under 90 min
WAV 48 kHz 24-bit (studio)2,304 kbps16.9 MB~60 minutesConvert before upload
WAV 44.1 kHz 16-bit mono706 kbps5.2 MB~198 minutesPlenty for speech
WAV 16 kHz 16-bit mono256 kbps1.9 MB~9 hoursWhat recognition uses
FLAC (lossless, compressed)~700 kbps~5 MB~200 minutesBest of both
MP3 128 kbps128 kbps0.9 MB~18 hoursLossy but ample

The practical takeaway: if your WAV runs past about 90 minutes, convert it to FLAC before uploading. FLAC is lossless — the samples come back bit-for-bit identical — but stores them in roughly half the space, which moves the ceiling from 99 minutes to over three hours. Converting to mono halves it again and costs nothing for speech, since recognition collapses the channels anyway. What is not worth doing is converting to MP3 out of caution: it does fit, but you are discarding data for a limit you had already cleared.

FAQ

Questions about WAV transcription

Convert your first WAV to text free

New accounts include about ten minutes of transcription — which for uncompressed audio is roughly a 100 MB file. Enough to run a real recording through and judge the result on your own audio.

Transcribe a WAV file