A standard stereo fold-down adds the center and surround channels back into left and right at 0.7071 gain, roughly 3 dB down.
Those coefficients were written into international recommendations decades before anyone sold a 3D plugin to bedroom producers, and they are still what a playback device reaches for when it has two speakers and a multichannel file. Height content gets folded in as well. You do not get a vote in any of it.
The ability to spread sources across a full sphere for the price of a decent reverb has turned that arithmetic into everybody's problem. It used to concern a few dozen dubbing stages. Now it concerns anyone who put a pad above the listener's head on a Tuesday afternoon.
Hard-panned material fares worse. A signal sitting fully left or fully right drops roughly 6 dB the moment both channels sum to a single one, which is what happens on a phone, a laptop, a single smart speaker, and most club systems below the main hangs.
So a balance that felt correct inside the sphere arrives front-loaded and level-shifted. Elements you pushed outward to make room now sit closer to the middle and quieter than the ones you left there deliberately, and the relationships you spent an afternoon setting are quietly rewritten.
There is a hint about all this buried in the delivery rules. Apple wants a separate stereo master, conformed and synced to the immersive file, rather than deriving one from it. Spotify, meanwhile, has no immersive tier at all. For most catalogs, the two-channel version is what the overwhelming majority of listening actually uses.
Reverb tails are the usual casualty. A tail spread wide and high sounds enormous in the render, and thin or hollow once summed, because most of what made it enormous was decorrelation rather than level. Granular ambience and modulated pads suffer for the same reason, since decorrelation is the entire mechanism.
Timing offsets do their own damage. A five millisecond difference between two channels cancels 100 Hz, and its odd multiples while reinforcing the even ones, producing a filtered, comb-shaped response that nobody chose and everybody hears.
The fix is not subtle, and it is not expensive. Audition your tails summed, not just your dry sources, because the dry signal usually survives the fold and the space around it usually does not.
Put the meter on the folded output rather than the wide bus, because what survives when both channels sum to one is the only thing in question here. Watching correlation on a 9.1.6 mix tells you very little, since the problem does not exist until the channels have been added together.
Loudness is the second check. Compare the integrated level of the immersive render against your stereo master, since a mix that leans on height and rear placement usually loses more energy in the fold than one built around the front, and that difference shows up as a quieter track in a playlist.
Then keep everything below roughly 100 Hz centered and mono. Bass placed off to one side or overhead gains almost nothing spatially, because low frequencies carry poor directional information anyway, and it is the content with the most to lose when phase relationships collapse into a single channel.
Those coefficients were written into international recommendations decades before anyone sold a 3D plugin to bedroom producers, and they are still what a playback device reaches for when it has two speakers and a multichannel file. Height content gets folded in as well. You do not get a vote in any of it.
The ability to spread sources across a full sphere for the price of a decent reverb has turned that arithmetic into everybody's problem. It used to concern a few dozen dubbing stages. Now it concerns anyone who put a pad above the listener's head on a Tuesday afternoon.
The fold-down math was settled long ago
The standard equations are blunt. Left output equals left at unity, plus center at 0.7071, plus left surround at 0.7071. Right works the same way. Everything you carefully placed behind the listener arrives in front, about 3 dB quieter than it was.Hard-panned material fares worse. A signal sitting fully left or fully right drops roughly 6 dB the moment both channels sum to a single one, which is what happens on a phone, a laptop, a single smart speaker, and most club systems below the main hangs.
So a balance that felt correct inside the sphere arrives front-loaded and level-shifted. Elements you pushed outward to make room now sit closer to the middle and quieter than the ones you left there deliberately, and the relationships you spent an afternoon setting are quietly rewritten.
There is a hint about all this buried in the delivery rules. Apple wants a separate stereo master, conformed and synced to the immersive file, rather than deriving one from it. Spotify, meanwhile, has no immersive tier at all. For most catalogs, the two-channel version is what the overwhelming majority of listening actually uses.
Decorrelated tails are the first thing to go
The 3D effect depends on differences between channels. A fold-down removes exactly those differences, which means the elements that sound most impressive in the sphere are the ones most likely to disappoint on the way out.Reverb tails are the usual casualty. A tail spread wide and high sounds enormous in the render, and thin or hollow once summed, because most of what made it enormous was decorrelation rather than level. Granular ambience and modulated pads suffer for the same reason, since decorrelation is the entire mechanism.
Timing offsets do their own damage. A five millisecond difference between two channels cancels 100 Hz, and its odd multiples while reinforcing the even ones, producing a filtered, comb-shaped response that nobody chose and everybody hears.
The fix is not subtle, and it is not expensive. Audition your tails summed, not just your dry sources, because the dry signal usually survives the fold and the space around it usually does not.
Two meters catch nearly all of it
A correlation meter reads +1 for identical channels, 0 for fully decorrelated material, and -1 for inverted polarity. Anything between +1 and 0 sums without trouble. Once the needle spends time below zero, a listener hearing the summed signal gets colored or severely attenuated audio, and that is the whole warning.Put the meter on the folded output rather than the wide bus, because what survives when both channels sum to one is the only thing in question here. Watching correlation on a 9.1.6 mix tells you very little, since the problem does not exist until the channels have been added together.
Loudness is the second check. Compare the integrated level of the immersive render against your stereo master, since a mix that leans on height and rear placement usually loses more energy in the fold than one built around the front, and that difference shows up as a quieter track in a playlist.
Then keep everything below roughly 100 Hz centered and mono. Bass placed off to one side or overhead gains almost nothing spatially, because low frequencies carry poor directional information anyway, and it is the content with the most to lose when phase relationships collapse into a single channel.
