What spatial audio is, and isn’t

01 · of 8

What does “spatial audio” actually mean?

It is an umbrella name for sound that seems to come from all around you. It is not one format.

One label, four different things underneath. None outranks the others.

In general use, it means sound beyond stereo. Google describes it as feeling that sounds come from all around you, not just left and right.

Apple uses Spatial Audio as the name of a feature. With supported headphones, it says, “the sound seems to come from all around you.”

YouTube uses the same words for first-order Ambisonics. Sony uses them for an object-based system.

The ITU’s Recommendations don’t use the name. They say “advanced sound system”.

Apple can also spatialise a plain stereo file. So “spatial audio” on a phone may come from a stereo master.

SoTreat “spatial audio” as a promise about the experience. Always ask which format is underneath.

02 · of 8

How do stereo, surround and immersive differ?

Stereo and surround sit on one ring at ear height. Immersive usually adds speakers above you.

TryStep from stereo to 7.1.4.Notice5.1 adds speakers around you. 7.1.4 adds a second ring, above.

The ITU lists them as separate kinds of system: two-channel stereo, multichannel stereo, meaning 5.1, and advanced sound systems.

Stereo is two speakers at 30° either side. 5.1 adds a centre, two surrounds and an optional LFE, all at ear height.

Height is what most people mean by immersive. An ITU report asks systems beyond 5.1 for sound “in all directions around the listener, including elevation.”

But its definition is wider. Stereo, 5.1 and headphones can carry an immersive programme too, with metadata.

Where every speaker goes is in Speaker layouts.

SoSurround adds speakers around you. Immersive adds height, or metadata that lets a renderer place sound in 3D.

03 · of 8

What is binaural, and what does head tracking add?

Binaural is a two-channel signal for headphones that carries the cues your ears use to place sound.

Turn your head

TryTurn 60° to the right in each mode.NoticeWorld-locked, the sound stays put in the room, so you hear it to your left. Head-locked, it follows your nose.

The EBU describes binaural as using the response of the human ear to give an immersive impression over two channels.

Those cues are an HRTF: how your body, head and outer ear filter a sound on its way in.

Apple says its headphone renderer makes “a binaural Spatial Audio mix”. Personalized Spatial Audio scans your ears and head, and Apple calls the result “a personalized HRTF”.

The ITU treats headphones as a full output, System Z. “Some programmes are even produced only for headphone playback.”

Head tracking changes the render as you turn. Apple’s Fixed setting is Spatial Audio without it.

SoBinaural is a headphone render built from ear cues. Head tracking decides whether the scene stays put when you turn.

04 · of 8

Channels, objects or scenes: how is it carried?

As speaker feeds, as separate sounds with position metadata, or as a sound field not tied to speakers.

Channel-based audio is mixed to a set number of channels, each for one speaker. That is a bed.

Object-based audio keeps sounds separate, with metadata a renderer reads at playback. Scene-based audio stores the whole sound field; Ambisonics is the example.

The ITU says the three can be combined in one programme, or used alone.

Go deeper in Beds, objects and metadata and Ambisonics.

SoA bed, a moving object and an Ambisonic scene are different kinds of signal. Each needs different handling.

05 · of 8

How does one master reach speakers and headphones?

A renderer reads the audio and its metadata, and builds a feed for whatever is playing it.

Two legitimate ways to reach every output. Neither is wrong.

The EBU describes a renderer as reading audio, metadata and the desired output, then generating the sounds the metadata describes.

Objects always need one. Channel-based audio may not, if the channels match the speakers.

Renderers differ. The EBU notes they “may produce subjectively different sounding outputs.” Check on more than one if you can.

Some tools render to a fixed 7.1.4 bus first, then make headphones and stereo from it. Orbit works this way.

SoYou make one master. Speakers and headphones each get their own render of it, so check on both.

06 · of 8

What isn’t spatial audio?

It isn’t one format, and it isn’t the same as binaural. An upmixed or widened stereo file isn’t an immersive mix either.

It isn’t one format or codec.: MPEG’s own white paper says there is no “one-size-fits-all” format for immersive audio.
It isn’t the same as binaural.: Binaural is one way to hear it, on headphones. The same mix plays on speakers too.
An upmix isn’t an immersive mix.: The ITU describes upmixing as generating the “missing” channels. Nothing new was mixed.
A widened stereo file isn’t one either.: Spatialising stereo at playback can sound wide, but the master is still two channels. This is our reading; no standard spells it out.
A 7.1.4 file isn’t an object mix.: A bed fixes every sound into its channels. Objects keep their positions as metadata.

SoAsk two questions of any “spatial” file. How many channels does it really have, and what metadata came with them?

07 · of 8

Which formats deliver it?

ADM BWF for masters, then delivery formats such as IAMF and MPEG-H 3D Audio.

Format	What it is for
ADM BWF	The production and exchange master. Apple Music asks for immersive masters in it.
IAMF	AOMedia’s open delivery format. Eclipsa Audio™ is based on it.
MPEG-H 3D Audio	A delivery format for channels, objects and Ambisonics. Sony’s 360 Reality Audio is built on it.

The details are in Reading an ADM BWF master and IAMF and Eclipsa Audio™.

SoMaster in ADM BWF, then deliver in the format each platform asks for.

08 · of 8

Where do I start?

Open one immersive master and switch between 7.1.4, binaural and stereo. You will hear the difference at once.

OrbitCheck an ADM BWF master on speakers, binaural, Apple Spatial Audio or stereo. Binaural, Apple Spatial Audio and stereo all come from one 7.1.4 bus, with head tracking from AirPods or your Mac’s camera.
Orbit ProEverything in Orbit, plus export from ADM BWF to IAMF and Eclipsa Audio™.
Orbit LiteA player for ADM BWF and IAMF files, on Mac, iPhone and iPad.
EsferaOur streaming platform: IAMF in the browser, rendered binaurally by OrbitCore, with nothing to install.
OSLSDescribe your own speaker rig in an open file, and Orbit outputs to it, up to 64 channels.

Coming soon from us: Cota, for loudness to ITU-R BS.1770-5 from stereo to 9.1.6. And Orbit Monitor, for headphone monitoring in your DAW.

Others are worth knowing too. Google’s free Eclipsa Audio™ plugin, still in beta, exports IAMF from your DAW. Your DAW may have its own renderer.

SoOpen one master in Orbit, switch between 7.1.4, binaural and stereo, and listen for what changes.

The words

Glossary

Advanced sound system: The ITU’s term for systems beyond 5.1, or any layout fed with channel, object or scene content plus metadata.
Immersive: Sound from all around you, usually including height.
3D audio: In MPEG-H, a setup with speakers above, and maybe below, ear level.
Surround: 5.1: three front speakers, two at the sides or rear, and an optional LFE, all at ear height.
Binaural: Two-channel audio for headphones that carries the cues your ears use to place sound.
HRTF: How your body, head and outer ear filter a sound arriving from one direction.
Head tracking: Changing a headphone render as your head turns, so sound can stay fixed in the room.
Channel-based: Audio mixed to fixed speaker feeds, such as 5.1 or 7.1.4. A bed.
Object-based: Separate sounds with metadata, such as position, that a renderer reads at playback.
Scene-based: A sound field stored as coefficient signals, not speaker feeds. Ambisonics is the example.
Renderer: The processor that turns audio and metadata into feeds for speakers or headphones.
Upmix: Making channels a mix never had, such as stereo to 5.1.

Sources