How do you compress a WAV file, and how much smaller does each option make it?

Why a WAV is as big as it is

A PCM WAV stores every sample as it is, so its size is just arithmetic: sample rate times bytes per sample times the number of sides, times the length, plus a header. Every file in this test matched that formula to the byte (the WAVs this site writes have a 44-byte header; the ones FFmpeg wrote had 78 to 114 bytes of header and tags).

So there are exactly four ways to make a WAV smaller as a WAV: fewer bits per sample, fewer samples per second, one side instead of two, or less length. The music itself does not matter: three minutes of silence is the same size as three minutes of orchestra. That is also why zipping does so little (see below), and why MP3, which spends bits only on what can be heard, gets so much smaller.

How we tested

• Music: 3 minutes (1:00 to 4:00) of Mozart's Symphony No. 40, first movement, played by the Musopen Symphony. It is in the public domain on Wikimedia Commons as a 48 kHz, 24-bit FLAC. FFmpeg 4.4.2 wrote it out as the four WAVs in the table above (soxr resampler, no dither).
• Browser: Chromium 153 driven by Playwright, opening a local copy of this site's WAV to MP3 page. One fresh tab per file; whole file, volume 100%.
• Score: signal-to-error ratio in dB against the 24-bit original (resampled to 44.1 kHz) after lining the files up. Higher is closer; it measures how far the waveform moved, not what you hear. Mono files are scored against the original's mono mix, so the score shows what the setting lost on top of going mono.
• Top frequency: the highest frequency where the file's average spectrum stays within 20 dB of the original's. The original reaches 21.3 kHz.
• Sizes are the exact bytes the browser downloaded. Times are on a server, not a phone.

Staying WAV: what each setting saves

The page always writes 16-bit PCM WAV. The first five rows are the bit-depth and sample-rate conversions people most often need; the rest start from the 16-bit 44.1 kHz file and go lower.

• 24-bit to 16-bit is the free one. It cut a third off and scored 73.6 dB; going to 44.1 kHz at the same time cut 39% off at 64.2 dB. The 96 kHz file lost 69% of its size for 62.6 dB. For comparison, FFmpeg's own 16-bit 44.1 kHz version of the same audio scored 79.9 dB: the browser's resampler and its rounding cost a few dB, but all of these are far above any MP3 in the next table.
• Mono and 22.05 kHz each halve the file, but they are not equal. Mono kept the full frequency range and 72.2 dB against the mono mix; 22.05 kHz kept both sides but nothing above 10.6 kHz and scored 41.9 dB. 16 kHz mono gave 18.1% at 7.7 kHz.
• Already 16-bit 44.1 kHz? WAV output does nothing. The file came back the same size, and 10,083,117 of its 15,876,000 samples were identical. The other 5,792,883, all positive values, came back 1 step lower: the page decodes to floating point and rounds back down without dither. It is not a bit-exact copy, so keep the original.

Going MP3: much smaller for less loss

• Size is bitrate times length. 320 kbps is 2.40 MB a minute whatever is in the file; 128 kbps is 0.96 MB. The WAV-only settings only get near this by throwing a lot away: the smallest WAV above, 16 kHz mono at 18.1%, sits between 320 kbps (22.7%) and 192 kbps (13.6%), and those MP3s keep both sides and sound up to 20.0 and 18.2 kHz where the WAV keeps one side up to 7.7 kHz.
• The MP3 scores are lower numbers, but they measure something different. MP3 removes what the encoder judges inaudible, which moves the waveform a lot; a lower sample rate removes only the top of the spectrum, which moves it little. 128 kbps stereo (15.2 dB) and 22.05 kHz WAV (41.9 dB) are not 27 dB apart in what you hear. Listen to both before you decide.
• A 24-bit or 96 kHz WAV does not make a better MP3. The 24-bit 48 kHz file at 192 kbps came out the same 4,320,862 bytes and the same 27.6 dB as the 16-bit file. We saw the same from 24-bit and 96 kHz FLAC.
• MP3 adds a little length: the files played 180.036 to 180.072 s for 180.000 s of WAV (encoder padding; the WAV outputs were exact to the sample, except the 48 and 96 kHz conversions, which came out one sample short at 7,937,999).

For the full set of low settings (64 kbps, 24 and 32 kHz, and what re-saving an MP3 costs) see how small an MP3 can get; if your WAV is a loud master, see how much headroom it needs before you convert.

Zipping a WAV, and lossless formats

"Compress" can also mean putting the file in a zip. We tried (deflate, level 9, Python's zipfile):

General-purpose compression finds almost nothing to repeat in audio samples, so a zip saves 5 to 11% and the recipient still has to unzip it. A lossless audio format like FLAC roughly halves a 16-bit file with no change to the samples, but this site cannot write FLAC; it can only open it. If you need lossless and small, FLAC is the right tool, just not this one.

What each option takes away

• 16-bit: the page decodes the file to floating point and writes 16-bit without dither. Against the 24-bit original, its 16-bit files scored 72.6 to 73.6 dB; FFmpeg's 16-bit version of the same audio scored 79.9 dB. Both are far above the 15 to 40 dB of the MP3s. If you will keep editing the file, keep the 24-bit original and convert a copy.
• A lower sample rate cuts the high frequencies: 32 kHz kept up to 15.3 kHz, 22.05 kHz up to 10.6 kHz, 16 kHz up to 7.7 kHz. The browser resampled the length correctly in every case (the 22.05 kHz files had exactly 3,969,000 samples for 180 s).
• Mono throws away the stereo spread. Mono mixes left and right into one signal, so whatever differs between them is lost. In this recording that difference carried 25.4% of the energy, and against the stereo original every mono file scored only 5.5 to 5.9 dB, the same as a perfect mono mix (5.9 dB). A recording that is already mono, or where both sides are the same, loses nothing here.
• Length is the one dial with no loss. Half the length is half the size at any setting. Type a start and end on the WAV to MP3 page, or cut the part you need first.

Steps

To keep it WAV (a 24-bit, 32-bit float or 96 kHz file):

1. Open WAV to MP3 and choose your WAV. The line under the name shows its sample rate, bit depth and mono or stereo.
2. Under Output format pick WAV (16-bit PCM), sample rate 44.1 kHz (or 48 kHz to change only the bit depth), and leave the mono/stereo choice on Same as file.
3. Press Convert, then Download. Title, artist and other LIST INFO tags are kept (see the page's FAQ).

To make it as small as possible:

1. On the same page pick MP3. For music, 192 kbps stereo (14%) kept 27.6 dB; 128 kbps (9%) kept 15.2 dB and sound up to 16.0 kHz.
2. For speech, choose Mono and 22.05 kHz first; the bitrate list then offers 48 kbps (3.4%). We only tested music, so listen to the result on a voice recording before you delete anything.
3. Press Convert and Download.

Questions

Can the page write 24-bit WAV, or FLAC?

No. Its WAV output is always 16-bit PCM, and it has no FLAC output. It opens 8 to 32-bit PCM, 32-bit float, A-law and μ-law WAVs, and FLAC on the audio cutter.

My WAV is already 16-bit 44.1 kHz stereo. What now?

WAV output cannot make it smaller, as the table shows. Your choices are mono or a lower sample rate (each 50%), a shorter file, or MP3 (23% at 320 kbps). For most uses MP3 at 192 or 320 kbps is the better trade.

Is the file uploaded to be compressed?

No. The page decodes and writes the file in your browser; see how to check whether an online tool uploads your file.