How can an MP3 throw away most of a song and still sound the same?

An MP3 can throw away most of a song's data and still sound like the original to most listeners because it mainly discards what human hearing is unlikely to notice. Compared to CD-quality digital audio, MP3 compression can commonly achieve a 75 to 95 percent reduction in size. Uncompressed CD audio runs at 1,411,200 bits per second, and MP3 was designed to encode that at 320 kbit/s or less. A commonly used rate of 128 kbit/s gives a compression ratio of 11 to 1, offering adequate quality in a small space.

The trick is a limit of human hearing called auditory masking: the perception of one sound is affected by the presence of another. Picture a cat scratching a post in a quiet room, which one listener first hears at 10 dB SPL. With a vacuum cleaner running, that same listener could not detect the scratching unless it reached at least 26 dB SPL. In the same way, a powerful spike at 1 kHz will tend to mask a lower-level tone at 1.1 kHz. That limit of hearing is what the MP3 algorithm takes advantage of, together with lossy compression, which reduces precision and partially discards data. The physicist Alfred Mayer had reported in 1894 that a tone could be made inaudible by another tone of lower frequency.

There is always a trade. The higher the bit rate, the larger the file and, generally, the closer it sounds to the original. But the combination of small size and acceptable fidelity led to a boom in sharing music online in the late 1990s, when bandwidth and storage were at a premium.