MP3 files: how it works, and why it is still everywhere

The consequence is generation loss. Every re-encode applies the model again to audio that has already been through it, and the second pass makes decisions based on damage rather than on the original signal. This is why converting an MP3 to a higher bitrate accomplishes nothing, and why the right workflow is always to keep a lossless master and encode from that when you need a new format.

MP3 lost on technical merit long ago — AAC beats it at the same bitrate, and Opus beats AAC, especially for speech at low rates. It won on ubiquity. Its patents expired in 2017, hardware decoders were built into a generation of cars, players and phones, and the result is a format that plays on things nothing else will touch. That, and nothing else, is the reason to convert to MP3 today.

Two practical notes. Bitrate: 192 kbps is transparent enough for most music on ordinary equipment, 320 kbps is the ceiling worth paying for, and 128 kbps or lower is fine for speech. And gapless playback: MP3 frames have fixed sizes, so encoders pad the start and end, which puts an audible gap between tracks that should run together. Live albums and DJ mixes suffer; formats designed later carry the padding information needed to remove it.