WAV, FLAC and MP3: What Is Actually Lost at Each Step

WAV: the editable source

WAV is a container used for PCM audio - which stands for Pulse-Code Modulation. A WAV file's bitrate is determined by its sample rate, bit depth, and channel count. The bitrate formula is: Sample Rate × Bit Depth × Channel Count.

Here's a concrete example: 44,100 Hz, 16-bit, stereo PCM produces a file with a 1,411.2 kbps bitrate. Drop down to mono, and the same parameters create a 705.6 kbps file, since there's just one sound channel.

So a WAV file is the exact digital representation of the original audio recording - the editable, untainted source. But that also means WAV files are quite large, since they store the full audio signal uncompressed.

FLAC: same audio, smaller file

FLAC is a different proposition. It's a lossless audio coding format: it allows the same LPCM audio data to be compressed significantly without losing information. FLAC does this by encoding the waveform in a compressed format while retaining the full digital fidelity of the original recording in the same LPCM form that WAV files use. The result is a far smaller file that thanks to the lossless compression mathematically represents the same audio signal in every way. Only the file size has changed.

The FLAC format was in 2024 formally specified by RFC 9639 from the RFC Editor. As the RFC puts it, FLAC is a lossless audio coding format specified "without losing information." That's not a limiting definition or technical specification, but a lossless encoding of the same LPCM audio data as a basic WAV file. But a CD-quality WAV file will be significantly larger than its FLAC equivalent, while containing the same audio data in the same form. The only thing "lost" is the file size, not the audio fidelity.

The Library of Congress describes FLAC as an open source bitstream encoding format designed for lossless compression of LPCM audio data, with default parameters tuned to CD-quality music data.

MP3: the irreversible step

The key step is the transition from FLAC or WAV to MP3. Converting a WAV or FLAC file to MP3 is where specific audio data is deleted and lost, permanently and irreversibly.

MP3 is a lossy compression format, just like JPG is a lossy compression format for photographs. So called because lossy compression deletes certain audio data. And unlike JPG, MP3 doesn't just delete less important audio data, but actually changes the way music sounds, using psychoacoustic artifacts to fill in the gaps.

Codec documentation tends to describe these "artifacts" as necessary approximations to make audibly transparent compression possible. But the net effect of those "necessary approximations" is a big reduction in the maximum audio quality, right down to the lowest end of the quality/sample rate/datastream that you found acceptable. MP3 at low bitrates will sound different from higher bitrate MP3.

In mathematical terms, encoding a WAV or FLAC audio file to MP3 is an irreversible step. The original audio data is deleted, with features that are difficult for humans to articulate - details, harmonics, texture, overtones - being mapped out by the encoder. The human brain fills in the blanks, which is why we need to instantiate "transparent" MP3 encoding separately for every single piece of music. Listen to a few MP3 files at high bitrate and you can hear what you're losing, particularly in terms of high frequencies, fullness, and space.

A product like this that's been decoded back from MP3 to WAV or FLAC has had audio quality permanently altered in ways that have no analogs in the original recording.

The original files may be bit-perfect copies of the recording, but digitizing them means picking and choosing aspects of the recording's sonic texture and cutting the rest, while also changing how you hear the music. Whether or not that shift in music making is positive depends on whether you see the loss of human ability to perceive differences as a good thing or a bad thing, and that, after all, is a matter of taste.

Re-encoding and recovery

One final point to note. Once audio has been encoded to MP3 and compressed, the information is gone. There is no such thing as converting an MP3 file back to FLAC or WAV to "recover" lost quality. The only change is from MP3 to another compressed MP3 version of perhaps slightly higher quality, but the audio data has still been lost. Worse, when lossy compressed to MP3, any further re-encoding further compounds the loss - with an accumulation of each format's typical compression-related distortions and errors - and always applied in a slightly different way every time. Absolutely don't keep your masters in MP3. Only ever use MP3 as a final delivery format. Or for specific ripping tools where absolute quality is not the priority.