MP3 vs WAV vs FLAC vs AAC: Which Audio Format Should You Use?

Digital audio starts as a long list of numbers: the height of the sound wave measured tens of thousands of times per second. Uncompressed formats such as WAV and AIFF simply write that list to disk. Nothing is removed and nothing is packed, which makes the files easy for any editor to read and fast to scrub through, but also large. A three-minute song stored this way takes around 30 MB, whatever it sounds like.

Lossless formats such as FLAC and ALAC pack the same list more cleverly, the way a ZIP file packs a document. When you play the file, the decoder rebuilds every sample exactly, so a FLAC decoded back to WAV is identical to the original down to the last bit. The saving depends on the music: quiet acoustic recordings shrink a lot, dense rock or noisy field recordings shrink less. Typical albums land at roughly half to three fifths of the WAV size.

Lossy formats such as MP3, AAC, Vorbis and Opus throw information away on purpose. An encoder uses a model of human hearing to drop sounds that are masked by louder ones nearby or that sit at frequencies you are unlikely to notice, then stores what is left. The result is five to ten times smaller than WAV. At a generous bitrate most people cannot pick it out from the source, but the removed detail is gone for good and cannot be restored by converting back.

The history explains a lot of the compatibility you see today. MP3 had a ten-year head start in consumer devices, WAV came bundled with Windows, FLAC grew up in the open-source world, and AAC rode in on the iPod and the iTunes Store. Open each row for the short version, including the four formats you will meet alongside the big four.

File size for uncompressed and lossy audio is simple arithmetic: bitrate times duration. WAV at 16-bit/44.1 kHz stereo runs at 1,411 kbps, which works out to about 10.6 MB per minute. Lossy formats use whatever bitrate you choose, so an MP3 at 320 kbps is about 2.4 MB per minute and at 128 kbps just under 1 MB. FLAC is the odd one out because its size depends on the music, so its numbers are always a typical range.

Those figures are for stereo. Mono halves them, which matters for voice recordings, and 24-bit/48 kHz WAV from a camera or a field recorder runs at about 17 MB per minute. As a quick sense of scale, a 45-minute album is roughly 480 MB as WAV, around 260 MB as FLAC, 108 MB as 320 kbps MP3 and 45 MB at 128 kbps.

Sample rate is how many times per second the wave is measured. CD audio uses 44.1 kHz, video and broadcast use 48 kHz, and studios sometimes record at 96 kHz. A sample rate can capture frequencies up to half its value, so 44.1 kHz already reaches about 22 kHz, past the roughly 20 kHz upper limit of human hearing. Higher rates help editing headroom more than they help playback.

Bit depth is how precisely each measurement is stored. 16-bit gives about 96 dB of dynamic range, which covers any listening room; 24-bit gives recording engineers extra room so a quiet take can be raised without hiss. Bit depth applies to PCM formats like WAV, AIFF, FLAC and ALAC. Lossy codecs do not store samples this way, which is why an MP3 has a bitrate but no bit depth.

Bitrate is how much data one second of audio uses, measured in kilobits per second. For uncompressed audio it follows from the other two numbers, multiplied by the channel count. For lossy formats it is the quality dial you set: 128 kbps is plenty for speech, 192 kbps is the usual middle ground for music, and 320 kbps is the MP3 ceiling. AAC reaches comparable quality at a somewhat lower bitrate than MP3.

Compatibility, not sound quality, is usually what forces a decision. MP3 is the one format that every car stereo, USB stick player, browser, upload form and email client handles. AAC is native on every Apple device and almost every phone, but some older car units and media players skip .m4a files. FLAC plays on Android, Windows 10 and later, and macOS, yet the Apple Music app still will not import it, which is why Apple users keep ALAC instead.

Production and distribution work the other way around: the destination wants the most complete version you have. Audio editors and DAWs read WAV and AIFF most reliably, and music distributors such as DistroKid ask for WAV or FLAC at 16- or 24-bit so the streaming services can make their own encodes. Podcast hosts and Apple Podcasts accept MP3 and AAC.

One rule covers most audio mistakes: never re-encode lossy to lossy twice. Every lossy encode throws away detail, and the next encoder cannot tell its own discards from the previous one's, so the losses stack. A 128 kbps MP3 converted to 256 kbps AAC is larger but not better, and after two or three rounds you start hearing swirly cymbals and smeared transients. Always encode from the best source you have, and keep that source.

Going from lossy to lossless has the opposite problem: it is safe but pointless for quality. Turning an MP3 into WAV or FLAC gives you a much bigger file that sounds exactly like the MP3. It is still the right move when an editor or a broadcast system only accepts WAV, as long as you know the quality ceiling was set by the MP3.

The conversions that pay off run downhill from a lossless master. Converting WAV to MP3 turns a 40 MB recording into something that fits under the 25 MB Gmail attachment limit, and a FLAC to MP3 copy puts a lossless album on a phone or a car stick. If you are on Apple gear and want to stay lossless, FLAC to ALAC keeps every sample and makes the files importable. Extracting sound from a video is its own case, covered in how to convert MP4 to MP3.

Each format family stores song details differently. MP3 uses ID3 tags, with ID3v2 frames for title, artist, album, track number, lyrics and an embedded cover image. FLAC and OGG use Vorbis comments, simple text fields plus a picture block for artwork. AAC in an .m4a file uses the MP4 metadata atoms that iTunes made popular, which is why album art and sort fields show up so cleanly on Apple devices.

WAV is the weak spot. Its RIFF INFO chunk stores a handful of text fields, cover art support is unofficial, and plenty of recorders and players write or read nothing at all. That is one more reason FLAC beats WAV for a music library: same audio, smaller files and proper tags. When you convert with the audio converter, Keep tags and cover art copies whatever the source holds into the fields the target format supports. If you need a transcript of spoken audio instead of a tidy library, see how to transcribe audio to text.

Find the row that matches what happens to the file next. When two rows apply, keep a lossless master and make the lossy copy from it, so you never have to choose between them.