The marketing word is "immersive." The mechanism is a feedback loop, and it is worth knowing because it explains why spatial audio needs sensors in your earbuds at all. The effect you are sold — sound that stays put in the world as you move your head — is impossible with a fixed stereo mix. The headphones have to know where your head is pointed and adjust the sound continuously to compensate.
Apple's grant US11202164B2, "Predictive head-tracked binaural audio rendering" (issued December 14, 2021), names all three pieces in its title. Head-tracked: the earbuds sense your head orientation. Binaural: the audio is rendered separately for each ear, with the tiny timing and tonal differences your brain uses to locate a sound. Predictive: the system anticipates your head's position slightly ahead, because if the re-render lags your motion, the illusion breaks and the sound smears.
Here is the loop, plainly. Sensors report that you turned your head 20 degrees left. The renderer recomputes the binaural mix so the on-screen voice, which should stay in front of you, now arrives as if from 20 degrees to your right — cancelling your motion. Do this fast enough, every frame, and the sound feels nailed to the world rather than glued to your ears. Miss the timing and you get the audio equivalent of motion sickness.
This is not one company's idea, which is the useful part. Sonos holds US12543015B2, "Spatial audio head tracker" (issued February 3, 2026), squarely on the head-tracking component. Nokia holds US12231871B2, "Spatial audio service" (issued February 18, 2025), on the service-and-rendering side. The CPC class H04S 7/304 — spatial audio with head-position compensation — recurs across all of them. Different assignees, same documented mechanism.
The myth worth puncturing: spatial audio is sometimes described as if the earbuds contain more or better speakers, or as if the sound is physically projected around you. They do not, and it is not. The drivers are ordinary; the magic is computational, and it depends entirely on the sensor-plus-render loop these patents describe. That is why spatial audio shows up as a software feature on hardware that already had the motion sensors — the speakers were never the point.
The precision note: each grant covers specific head-tracking or rendering methods, and the timing-prediction details are where the real engineering differences live. But the shared CPC class and the parallel filings tell you this is established, multi-vendor art. When the voice stays put as you turn your head, you are hearing a prediction loop run correctly — not a new kind of speaker.