How AAC Works
AAC (Advanced Audio Coding) is a lossy format built on the same core idea as MP3: use a model of human hearing to decide which parts of a sound can be discarded without a listener noticing, then spend the available bits only on what remains. It was designed from the start as MP3's successor, and it is better at the job because it fixes several structural limitations that were baked into the older format.
At its heart AAC uses a modified discrete cosine transform (MDCT) to convert short blocks of the waveform into frequency components, which a psychoacoustic model then quantizes more coarsely wherever the ear is least sensitive or where louder sounds mask quieter ones. MP3 did this too, but AAC works with a purer, more flexible filter bank and can switch between long blocks (efficient for steady tones) and short blocks (which avoid the smearing artifact called pre-echo on sharp transients like a drum hit) far more cleanly.
Several specific improvements add up to AAC's edge at any given bitrate. It supports sampling rates up to 96 kHz and far more than MP3's two channels, so it handles surround sound natively. It uses temporal noise shaping to keep quantization noise from spilling across sharp attacks, and joint stereo techniques that store the difference between left and right channels rather than each in full. The practical result is that AAC at 128 kbps generally sounds as good as MP3 at 192 kbps, and the gap widens at lower bitrates where AAC's efficiency matters most.
AAC is not a single setting but a family of profiles. Plain AAC-LC (Low Complexity) is the workhorse used by most streaming and downloads. HE-AAC (High Efficiency) adds spectral band replication to reconstruct high frequencies cheaply, making it excellent for low-bitrate streaming and digital radio, and HE-AAC v2 layers in parametric stereo for even lower bitrates. On disk the audio almost always lives inside an MP4-family container, which is why you see it as .m4a, .aac, or as the audio track inside an .mp4 video.
History and Standardization
AAC was developed in the 1990s by a group including Fraunhofer, Dolby, AT&T, Sony, and Nokia, several of the same organizations behind MP3, and was first standardized in 1997 as part of MPEG-2 (ISO/IEC 13818-7). It was later folded into the broader MPEG-4 specification (ISO/IEC 14496-3), which is where the additional profiles like HE-AAC were defined. It was engineered deliberately as "the successor to MP3," not a competitor to it.
What turned AAC from a strong standard into a dominant one was Apple's decision to build the iTunes Store and iPod around it in the early 2000s. Choosing AAC over MP3 for a platform that sold billions of songs gave the format enormous reach, and it became the default encoding for Apple Music, the iOS ecosystem, and much of the mobile world. YouTube, most broadcast and streaming platforms, and the majority of modern devices adopted it as well.
Unlike MP3, whose core patents have now expired, AAC remains patent-licensed, with fees administered through pools such as Via Licensing. In practice this is invisible to end users because the licensing is handled by the device and software makers, but it is one reason the fully open Opus codec was created as a royalty-free alternative for the web. It has not dislodged AAC from the entrenched Apple and broadcast ecosystems, where AAC's compatibility is decisive.
AAC Compared to MP3 and Opus
Against MP3, AAC wins on nearly every technical measure: better sound at the same bitrate, cleaner handling of transients, native multichannel support, and far better behaviour at low bitrates. The only place MP3 still leads is raw universality, a decades-old device with no AAC decoder will still play an MP3, and MP3 is now royalty-free while AAC is not. For anything made in the last fifteen years, AAC is the better choice.
Against Opus, the modern open codec, AAC generally loses on pure quality-per-byte, especially at very low bitrates and for speech, where Opus is exceptional. Opus is also free of licensing. What keeps AAC ahead in practice is hardware and platform support: it is decoded natively by essentially every phone, browser, TV, and car stereo, whereas Opus support, though now broad on the web, is still patchier on older consumer hardware.
One property AAC shares with MP3 is worth remembering: it is lossy, so the data its psychoacoustic model discards is gone permanently. Converting an AAC file to a lossless format like FLAC only wraps a larger container around the already-reduced audio, it does not restore quality, and re-encoding AAC to another lossy format compounds the loss. Editing and archival workflows therefore keep a lossless master and treat AAC purely as a delivery format produced at the final step.
