Apple requires a separate stereo master alongside every Dolby Atmos delivery, conformed and synced to within 50 milliseconds of the immersive version. Not derived from it. Delivered next to it, as its own file.
That single rule reshapes how you should think about animated panning. You get one stereo source and one immersive source per track, and the platform will not build the stereo from your spatial mix. Whatever rhythmic movement you programmed lives in one of those two files unless you deliberately put a version of it in the other.
Listeners can switch between the two on the fly, which is where the difference becomes obvious to anyone paying attention. A part built on tempo-locked pan moves carried by object metadata can be the hook of the immersive version and completely absent from the stereo one. Same song, two different arrangements, and only one of them got your attention.
Front and back balance is the third, pushing the remaining energy toward the front three speakers or toward the sides. If your pattern spends half its time behind the listener, that control is deciding a meaningful part of what the fold-down sounds like, whether or not you have looked at it.
The default 5.1 path applies fixed reductions on the way down, with the center dropped around 3 dB, the surrounds around 6 dB, and the LFE somewhere between 9 and 12 dB. Height trim matters most for anything you animated across the ceiling, because that is the material with nowhere to go once the overheads disappear. Those defaults are a starting point rather than a verdict, and adjusting them by hand is the only way to keep a specific move intact.
Rhythmic panning makes this harder to spot than a static placement does, because the phase relationship changes on every step. A pattern that sounds fine folded down at bar nine can go hollow at bar seventeen, purely because the source has moved to a position where it fights something else.
There is a shortcut worth knowing. Rendering straight to 2.0 rather than passing through an intermediate 5.1 stage gives a more transparent and more content-aware result. If your stereo check sounds worse than the material deserves, the routing may be the reason rather than the mix. Audition the re-render while you build the pattern, not after you have committed to it, since a move that collapses is easier to reshape than to rescue.
The file itself is a broadcast wave carrying audio definition model metadata, 24-bit and 48 kHz. The technical targets are equally specific. Integrated loudness should not exceed -18 LKFS measured to ITU-R BS.1770-4, and true peak should stay under -1 dBTP. Peak limiting indicated for binaural headphone playback should not exceed 3 dB, which quietly rewards a mix with some headroom left in it.
Sync tolerances are tight too. The two versions must land within 50 milliseconds of each other in duration, and track boundaries need to align within half a frame, about a thousand samples at 48 kHz. No extra silence at the end of the immersive version compared to the stereo one.
None of that stops you writing an aggressive pan pattern. It does mean the decision about how a spatial mix folds down to stereo is one you make twice, once in the trim controls and once in the stereo session where you decide whether the idea gets a two-speaker equivalent at all.
Sometimes it does not need one. A subtle ceiling flutter can be an immersive-only detail, and the stereo mix loses nothing. Other times the pattern is the arrangement, and shipping a stereo master without it means most of your audience hears a different song from the one you finished.
That single rule reshapes how you should think about animated panning. You get one stereo source and one immersive source per track, and the platform will not build the stereo from your spatial mix. Whatever rhythmic movement you programmed lives in one of those two files unless you deliberately put a version of it in the other.
Listeners can switch between the two on the fly, which is where the difference becomes obvious to anyone paying attention. A part built on tempo-locked pan moves carried by object metadata can be the hook of the immersive version and completely absent from the stereo one. Same song, two different arrangements, and only one of them got your attention.
Trim controls decide what survives fewer speakers
The renderer gives you three levers for this, and most people never touch them. Surround trim sets how much energy from the side and rear speakers carries into formats that have fewer of them. Height trim does the same job for the overhead information.Front and back balance is the third, pushing the remaining energy toward the front three speakers or toward the sides. If your pattern spends half its time behind the listener, that control is deciding a meaningful part of what the fold-down sounds like, whether or not you have looked at it.
The default 5.1 path applies fixed reductions on the way down, with the center dropped around 3 dB, the surrounds around 6 dB, and the LFE somewhere between 9 and 12 dB. Height trim matters most for anything you animated across the ceiling, because that is the material with nowhere to go once the overheads disappear. Those defaults are a starting point rather than a verdict, and adjusting them by hand is the only way to keep a specific move intact.
Opposite-side elements can cancel when they sum
Folding a wide spatial image into two speakers means content that used to be separate now shares a channel. When those elements are correlated, you get cancellations and summations that were never audible in the immersive version. Distorted guitars sitting in front-wide positions are a known offender.Rhythmic panning makes this harder to spot than a static placement does, because the phase relationship changes on every step. A pattern that sounds fine folded down at bar nine can go hollow at bar seventeen, purely because the source has moved to a position where it fights something else.
There is a shortcut worth knowing. Rendering straight to 2.0 rather than passing through an intermediate 5.1 stage gives a more transparent and more content-aware result. If your stereo check sounds worse than the material deserves, the routing may be the reason rather than the mix. Audition the re-render while you build the pattern, not after you have committed to it, since a move that collapses is easier to reshape than to rescue.
The stereo master is a separate piece of work
Apple's rules here are strict and worth reading before you plan the session. Upmixing a stereo release into Atmos is not allowed. Neither is extracting stems from a stereo release. The immersive version has to come from multitracks or from stems built out of multitracks.The file itself is a broadcast wave carrying audio definition model metadata, 24-bit and 48 kHz. The technical targets are equally specific. Integrated loudness should not exceed -18 LKFS measured to ITU-R BS.1770-4, and true peak should stay under -1 dBTP. Peak limiting indicated for binaural headphone playback should not exceed 3 dB, which quietly rewards a mix with some headroom left in it.
Sync tolerances are tight too. The two versions must land within 50 milliseconds of each other in duration, and track boundaries need to align within half a frame, about a thousand samples at 48 kHz. No extra silence at the end of the immersive version compared to the stereo one.
None of that stops you writing an aggressive pan pattern. It does mean the decision about how a spatial mix folds down to stereo is one you make twice, once in the trim controls and once in the stereo session where you decide whether the idea gets a two-speaker equivalent at all.
Sometimes it does not need one. A subtle ceiling flutter can be an immersive-only detail, and the stereo mix loses nothing. Other times the pattern is the arrangement, and shipping a stereo master without it means most of your audience hears a different song from the one you finished.
