Is a WAV more accurate than an MP3?

Slightly, and only when the audio is already difficult. Compression removes what the ear can't perceive, and speech recognition doesn't need it either.

Should I compress it before uploading?

It isn't necessary. We support up to 50 GB uncompressed; converting beforehand only adds time and one more step where something can go wrong.

My WAV is over 50 GB. What should I do?

That's about fifteen hours in stereo. You can split it by session, or convert that file to MP3: at that length, the quality difference is irrelevant.

Does it support 24-bit or 96 kHz?

Yes. A higher rate doesn't improve accuracy, but it doesn't reduce it either.

Is the transcript more accurate with a WAV?

The result is practically identical to a standard-quality MP3. The difference between the two formats lies in frequencies the voice doesn't use, so accuracy isn't affected.

How long does an hour-long WAV take to upload?

It depends on your connection: about 600 MB can take several minutes over a typical home upload. If you're in a hurry, convert it to MP3 first and it will be around 30 MB.

No account required

Turn a WAV file into text

Uncompressed audio, transcribed exactly as it was recorded. File size isn't a problem: we support up to 50 GB.

Drag your file in here

MP3, WAV, M4A, MP4, MOV… and any other audio or video, whatever the size

Up to 50 GB per file: an hour at 48 kHz fits with room to spare
No prior conversion or quality loss
Compatible with any sample rate and bit depth your recorder uses

WAV

About this tool

WAV is an uncompressed format: it preserves the audio signal exactly as captured by the microphone. It's the best possible starting material for speech recognition, though also the heaviest.

An hour in stereo at 48 kHz takes up around 600 MB, sixty times more than that same hour as an MP3. It's an upload limit, not a transcription one: we support files of up to 50 GB with nothing to compress beforehand.

Features

Compatible with any sample rate: 44.1 kHz or 48, 16 or 24 bit, mono or stereo.
Files of up to 50 GB, equivalent to several hours uncompressed.
No re-encoding: it transcribes exactly what your device recorded.
Multi-microphone recordings are correctly identified by voice.

How it works

1
Upload the WAV
It transfers directly from your browser to storage.
2
We transcribe it
Just as fast as any other format: what matters is duration, not file size.
3
Download the text
With timestamps, identified voices and a summary if you requested one.

Use cases

Field recordings

Portable recorders generate WAV by default: the right file for transcribing.

Studio sessions

The unprocessed track transcribes just as well as the mastered one, often better.

Historical archives

Old material kept as WAV can be transcribed without prior conversion.

In detail

Why a WAV file takes up so much space

Because it applies no compression at all. Every audio sample is stored exactly as captured by the microphone: 44,100 samples per second, two channels, two bytes per sample. That works out to ten megabytes per minute, six hundred per hour, and a full day of recording can approach five gigabytes.

In exchange, no data is lost, which matters when the recording is an original destined for archiving or further editing. Transcription doesn't strictly require this — the result is indistinguishable from a good-quality MP3 — but it doesn't cause any issue either.

How to upload a multi-gigabyte file without trouble

The per-file limit is 50 GB, and the upload goes directly from your browser to storage. Even so, a one-gigabyte file can take several minutes over a home connection, so there are two alternatives.

First: trim anything you won't need beforehand. Second: convert it to a 96 kbps MP3, which reduces the file to 5% of its original size with an identical transcript. Both options are available on this site, free of charge, without moving the file off your device.

Frequently asked questions

Need to transcribe a video?

Upload it and get the text with timestamps, each voice identified and a summary. No account needed to start.

Transcribe my recording See pricing