WAV or MP3: which file to release, and which to keep

What a WAV and an MP3 are

A WAV is a container. The Library of Congress, which keeps a register of digital file formats, describes WAVE in its WAVE format entry as a “wrapper file format”, and says “One common bitstream encoding is LPCM (Linear Pulse Code Modulation)”. It lists others, such as µ-Law, A-Law and ADPCM, and says the fidelity “Varies according to encoding”. The WAV your DAW exports is PCM: the samples it rendered, stored as they are. Nothing is thrown away, so a PCM WAV is lossless.

An MP3 is lossy. The same register’s MP3 format entry says its compression uses “perceptual coding”, built on models of hearing “that permit the codec to discard or reduce the precision of audio components that are less audible to human hearing.” DistroKid’s help on lossless formats puts it more plainly: “some of the audio data is removed to make the file size smaller.” What the encoder removed doesn’t come back, so converting an MP3 to a WAV gives you a bigger file of the same audio, not the original.

FLAC and Apple’s ALAC compress without losing anything. Apple’s Music user guide says ALAC “makes files smaller while preserving all of the original data in the file”, and DistroKid calls FLAC lossless, with the “same audio quality as WAV” in smaller files.

Whether you can hear a 320 kbps MP3 against its WAV depends on the song, the encoder and the listener, and we make no claim either way. To try it on your own song, load both files into our free Compare masters tool, which plays two files in sync at matched loudness. What isn’t in doubt is that the MP3’s missing data stays missing when the file is encoded again.

How big each file is

A WAV’s size is arithmetic: sample rate × bit depth × channels × seconds gives the bits, and eight bits make a byte. For a three-minute stereo song at CD quality, which is 16-bit and 44.1 kHz:

44,100 × 16 × 2 × 180 = 254,016,000 bits, which is 31,752,000 bytes, about 31.8 MB.

An MP3’s size comes from its bit rate instead. At 320 kbps, which is 320,000 bits a second:

320,000 × 180 = 57,600,000 bits, which is 7,200,000 bytes, about 7.2 MB.

Scrolls sideways

So the 16-bit WAV is about 4.4 times the size of the 320 kbps MP3 (1,411,200 ÷ 320,000), and a 24-bit 48 kHz WAV about 7.2 times. The Library of Congress gives the same kind of figure for a 128 kbps MP3, “about 1/11 the size” of CD-quality audio. Real files come out slightly larger, because each one also carries a header and any tags: we rendered three minutes in each format, and the WAVs came out 78 to 102 bytes over the sums and the 320 kbps MP3 about 2.5 KB over. MB here means a million bytes.

What distributors ask for

DistroKid’s page on audio formats says uploads “should be WAV, MP3, M4A, FLAC, AIFF, or Windows Media (WMA)”, and “If you’re sending a WAV, 16-bit, 44.1 kHz WAV is typical, but pretty much any specification works.” Its release checklist asks for a “Final mastered audio file (WAV or FLAC recommended)”, and its help on lossless formats says to upload an MP3 only “if you are comfortable with the sound quality of the MP3”. On Apple Music’s badges, DistroKid says the Lossless badge comes from uploading “a lossless audio file format” (FLAC, ALAC, WAV or AIFF), and that a Hi-Res file “must be at least 24-bit and have a sample rate above 48 kHz.”

TuneCore’s upload help recommends “24 bit (sample size) with 192 kHz (sample rate) .WAV files, but 16 bit (sample size) with 44.1 kHz (sample rate) .WAV files are also accepted.” It says “Your audio must meet the format specifications below to go live in the digital stores”, and those two are the only ones it lists, so a 24-bit WAV at 44.1, 48 or 96 kHz is neither. Unless your master is 24-bit at 192 kHz, send TuneCore the 16-bit 44.1 kHz WAV. The page gives 1411 kbps as the bit rate for both; that is the 16-bit 44.1 kHz figure from the sum above, since a 24-bit 192 kHz stereo file runs at 9,216,000 bits a second. If your master is at 44.1 or 48 kHz, don’t convert it up to 192 kHz to match. Apple’s guide for mastering engineers, Apple Digital Masters, says “Upsampling won’t recover or add information to an audio file.”

CD Baby’s upload help asks for “WAV or FLAC (preferred); MP3 accepted but not recommended”, in stereo at 44.1 kHz and 16-bit. It says a WAV’s file details should read “1,411 Kbps and 44.100 kHz”, and that “A common mismatch is 24-bit / 2,304 Kbps at 48.000 kHz”, which “needs to be converted down to 16-bit / 1,411 Kbps at 44.100 kHz”. It explains the MP3 advice too: “When we deliver your files to our digital partners, we encode them into the various formats each platform requires. MP3 is already a lower-quality, compressed format — so any quality loss it has is compounded by the encoding process.”

These pages were read on 3 November 2026. Check your distributor’s own page before you upload, because requirements change.

What Spotify and Apple Music stream

Neither service streams a WAV. What listeners hear is a copy encoded from your upload. Spotify’s loudness help asks for true peaks below −1 dB, which it says “is best for lossy formats (Ogg/Vorbis and AAC) and makes sure no extra distortion’s introduced in the transcoding process.” Its audio quality page and Apple’s Music user guide list what each listener gets:

Scrolls sideways

Apple Digital Masters says Apple chose “to standardize on AAC instead of the more popular MP3 format”, and that “Apple Music delivers lossless versions of these masters to our subscribers using the Apple Lossless Audio CODEC (ALAC.)” It asks for “only 24-bit sources” and “the highest-resolution master file possible”.

MP3 isn’t among the formats either service names for streaming. If you upload an MP3, the lossy streams are encoded again from it, by the distributor or the service, which is the compounded loss CD Baby describes. Our Spotify mastering guide covers what Spotify publishes about loudness.

When an MP3 is the right file

An MP3 is the practical choice wherever a small file is what you need:

• Sending a demo or a rough mix to a bandmate, a producer or a booker, by email or a message.
• Sharing a preview before the release, as a file that’s quick to send.
• Listening copies for your phone or the car, so you can check the song away from the studio.
• Anywhere that asks for MP3, or has an upload limit a WAV won’t fit.

It isn’t the file for your distributor, your archive or the mix you send to mastering. For mastering, send the WAV or FLAC of your mix. You can upload an MP3 here, but our free mix check flags it as a lossy source: “Encoding has already removed detail, and mastering makes what is left louder.” Our guide to exporting a mix covers the sample rate, bit depth and dither settings.

Keep the 24-bit WAV

The 24-bit WAV at your mix’s own sample rate is the master. Every other file can be made from it: a 16-bit WAV for CD Baby, TuneCore or a CD, an MP3 for a demo, whatever a service asks for later. None of them can be turned back into it. Apple asks engineers not to “bit-pad” or recapture 16-bit files in 24-bit, and making a 16-bit file takes dither, which Apple describes as “a technique used when reducing a file’s bit depth (for example, from 24-bit to 16-bit)”.

It’s also where the high-resolution options start, though 24-bit alone doesn’t always qualify. DistroKid’s Hi-Res badge on Apple Music needs at least 24-bit and a sample rate above 48 kHz, so a 24-bit WAV at 44.1 or 48 kHz doesn’t meet it. TuneCore recommends 24-bit, 192 kHz, and Apple’s guide asks for 24-bit sources. Keep the 16-bit 44.1 kHz WAV beside it for the distributors that ask for that, and keep copies of both in more than one place, not only on one laptop or with one service, ours included.

Scrolls sideways

The files that come with a master here

Every master download on Bounce Sessions, bought for $4.99 a track or covered by Unlimited at $15 a month or $99 a year, includes all of these:

• A 24-bit WAV at your mix’s own sample rate. We take mixes from 44.1 to 96 kHz, so a 96 kHz mix gets a 96 kHz master.
• A 16-bit, 44.1 kHz WAV, made from the master with dither.
• An MP3 at 320 kbps, encoded from the master itself, not from either WAV.

Buying a track covers every file of that master, so downloading the MP3 after the WAV costs nothing more.

An MP3 can peak higher than the master it was made from, which is why Spotify asks for headroom before encoding. On the default ceiling of −1.0 dBTP, we decode the MP3 after encoding and measure its true peak; if it would clip, it is encoded again slightly quieter, up to twice. If it is still above 0.0 dBTP after that, the job fails its final check and nothing is delivered. If you choose the −0.1 dBTP ceiling instead, which suits a WAV nobody re-encodes, the MP3 is delivered as encoded and its peak is measured, not held. The upload screen says so beside the choice.

The free preview plays the whole master against your mix, each at its own level, with no account needed. The master you hear is that same 320 kbps MP3, and your mix plays as a 128 kbps MP3 made from your upload.