Channels versus objects
A stereo mix is channel-based, and so is traditional surround sound — 5.1 and 7.1 are the same idea with more speakers. You decide how much of each sound goes to each speaker, and that decision is baked into the file. It assumes speakers in known positions, and if the listener has a different arrangement, the mix does not know and cannot adapt.
Dolby Atmos is object-based. Instead of assigning a sound to a channel, you give it a position in three-dimensional space — including height — and the file stores the audio and the position separately. The playback device then renders those positions for whatever it actually has: a cinema with sixty speakers, a soundbar, a pair of earbuds.
That is the entire idea, and everything else follows from it. An Atmos mix is not a mix for a specific speaker layout. It is a description of where things are, plus instructions the renderer follows.
In practice an Atmos music mix is a hybrid: a 7.1.2 channel-based “bed” for things that do not need to move, plus up to 118 objects for things that do.
How you are almost certainly hearing it
Here is the thing that gets lost in the marketing: the overwhelming majority of Atmos music listening happens on headphones, through a binaural render.
Binaural rendering simulates what your two ears would hear if the sounds really were at those positions, using a head-related transfer function — a model of how your head, torso and the shape of your ears alter sound arriving from different directions. Your brain reads those alterations as direction, and the sound appears to come from outside your head rather than from between your ears.
This works, and it works on completely ordinary headphones. There is no special hardware required, and headphones marketed as “spatial audio ready” are selling you something the format does not need. Head tracking — where the image stays put as you turn your head — is a genuine hardware feature, and it is an enhancement rather than a requirement.
The limitation is that the transfer function used is a generic one, and everyone’s ears are different. Some people get a vivid, convincing externalised image. Some people hear very little difference at all. Neither is doing it wrong, and this variability is a large part of why opinions about spatial music diverge so sharply.
What making one involves
You need the individual elements — the multitrack or a stem set. You cannot make a genuine Atmos mix from a stereo master, because there are no separate sources to place. Anything claiming to do so is applying a widening effect, which is a different and much smaller thing. Where the multitrack is genuinely gone, separation can produce sources to place — the result inherits whatever the separation left behind, which is a real cost and worth naming rather than hiding.
The workflow, in outline:
- Route each element to a bed channel or an object. Things that anchor the record — kick, bass, lead vocal — usually go to the bed and stay put.
- Position the objects, and automate the positions where movement serves the song. The failure mode here is obvious and common: things circling the listener because the format allows it.
- Monitor the binaural render constantly, because that is what listeners will hear. A mix checked only on a speaker array is a mix optimised for the room almost nobody has.
- Check the stereo fold-down. Platforms derive one, and the majority of plays are still stereo. A spatial mix that collapses badly in stereo has traded the audience it has for the one it wants — a mono-compatibility check is the fastest way to find out what disappears when it collapses.
The deliverable is a file format
An Atmos music deliverable is an ADM BWF file — a Broadcast Wave containing all the audio alongside metadata describing every object and its movement over time. It is large, it is precisely specified, and platforms validate it on ingest.
This is worth being clear about, because “we support Atmos” can mean two very different things. It might mean a tool produces a binaural headphone render that sounds spatial — useful, and not an Atmos deliverable. Or it might mean it produces a valid ADM BWF that a distributor will accept. The second is a professional obligation with a validation toolchain behind it.
If you intend to release in Atmos, confirm which one you are getting before you build a schedule around it.
Atmos loudness is different, and it matters
Atmos music is delivered at a target of −18 LUFS integrated, which is quieter than the −14 LUFS most people aim at for stereo streaming. That is a separate target for a separate deliverable — do not carry a stereo figure across, and do not carry this one back. What LUFS actually measures covers the measurement itself, and the Apple Music page covers how the stereo release is levelled on the platform where most Atmos listening happens.
Consequently, on a platform that offers both, the Atmos version can play noticeably quieter than the stereo one — and listeners comparing the two will usually prefer the louder, regardless of which is better. This is the same trap as any unmatched comparison, arriving in a new place.
There is also a limiter to consider: the Atmos renderer applies its own, so a mix that arrives already squashed gets processed twice. Leave headroom and let the format do its job — the same reason a mix bounced for mastering should arrive without a limiter on the bus.
Is it worth doing?
Honestly, it depends, and the arguments on both sides are real.
For: platforms promote spatial releases, which is a discovery advantage that has nothing to do with the audio. Some material genuinely benefits — dense arrangements gain separation, ambient and orchestral music gains space, and a record with a lot of layers can suddenly be heard as layers. And it is an interesting creative constraint.
Against: it is a second full mix, not a conversion. The listening experience varies with each listener’s ears, so you cannot be confident in what they hear. The loudness difference works against you in casual comparison. And a great stereo mix beats a mediocre spatial one every time, on every system.
The clearest advice: if your stereo mix is not finished and excellent, do that first — the stereo release is what the overwhelming majority of plays will be, and it is the version worth spending on. A spatial version of a record that does not work is a spatial version of a record that does not work.
Spatial audio, Atmos, and other words
These terms get used interchangeably and are not the same thing. Spatial audio is the general category — any technique for placing sound in three dimensions. Dolby Atmos is one specific object-based format, owned by Dolby, with its own specification, tools and validation. Binaural is a rendering technique for headphones, used by Atmos and by plenty of things that are not Atmos. Ambisonics is a different spatial representation entirely, common in VR and broadcast.
A product can be spatial without being Atmos, and binaural without being either. When a tool says it does “spatial audio”, the useful question is which of these it means and what file it hands you at the end.
Keep reading