Fixed vs Head-Tracked Spatial Audio

What fixed rendering does

A fixed mode can still apply HRTF, virtual-speaker, and room processing. The soundstage may appear in front of the listener, but the entire field generally follows the head. This can be comfortable for walking, reclining, or listening without a screen because the acoustic perspective remains centered on the listener.

Fixed is not the same as off. All the spectral and reflection processing that moves the image away from the earcups is still running; only the orientation input is absent. For a listener sitting still and facing forward, the two modes can be difficult to distinguish, because there is no head movement for tracking to respond to.

What tracking adds

Compatible headphones provide orientation data from motion sensors. The renderer rotates the virtual scene in the opposite direction of the head movement. If the listener turns right, a source fixed in front shifts toward the left ear, just as a real monitor would.

Apple exposes Off, Fixed, and Head Tracked choices for supported AirPods content. Availability still depends on device, operating system, app, and content, so support information should always be checked against Apple’s current documentation.

Beyond stability, tracking supplies the active test described in localisation research: it lets the auditory system change its own viewpoint and check whether the scene behaves like a real one. That is why tracking often improves front-back separation and perceived distance even for listeners who were not consciously moving their heads much.

Latency, drift, and recentering

Three implementation details determine whether tracking feels like a room or like a gadget. The first is latency between head movement and the corresponding change in the rendered signal. Small delays are tolerable, but as the lag grows the scene starts to feel like it is being dragged behind the head rather than staying put.

The second is drift. Orientation sensors accumulate error, so most systems slowly re-establish which direction counts as forward. The third is recentering policy: if you turn and hold that position, the system must decide whether the scene stays where it was or gradually treats your new orientation as centre. Holding indefinitely is geometrically correct but uncomfortable in practice, so most implementations recentre over time. Neither choice is wrong, but the transition should be slow enough that you never notice it happening.

If tracking sounds unnatural, these are the usual causes rather than the HRTF or the room model. Audible stepping suggests a low update rate; a scene that swims after you settle suggests aggressive recentering; a scene that lags suggests transport or processing delay.

Choose by situation

Tracking works especially well for a movie tied to a display, a game world, or a virtual pair of speakers. Fixed processing can feel better when moving around, listening casually, or when a screen anchor would be arbitrary. Some listeners also prefer a lower tracking strength for music because constant recentering can draw attention to the technology.

Posture is the practical deciding factor. Anchoring a scene to a fixed forward direction only makes sense if there is a meaningful forward direction. Working at a desk in front of a display, there clearly is. Walking, lying down, or listening on public transport, there is not, and a scene anchored to an arbitrary heading will behave oddly as you move.

• Desk work with a display: tracking, anchored to the screen.
• Films and games: tracking, so dialogue and effects stay with the picture.
• Music with eyes closed: either, often a matter of taste.
• Walking or commuting: fixed, since there is no stable reference direction.
• Critical listening to a stereo master: bypass entirely.

Evaluate the anchor, not the spectacle

Turn the head only ten or fifteen degrees. The scene should remain stable without a delayed jump, audible pitch change, or obvious level pumping. Return to center and check that the center image settles naturally. Good tracking disappears into the listening experience instead of behaving like a demo effect.

Large, fast head turns are a poor test. They exaggerate every implementation artefact and correspond to nothing you do while listening. The movements that matter are the small, near-continuous ones people make without noticing, and a system that handles those cleanly will be comfortable for hours.