Case Study — Immersive Music Production & Post

In brief

The hub for the music cases: the shared studio brief and the vocabulary of spatial creation.
Six axes, each with a technical handle: position, depth, size, trajectory, envelopment, density.
Cross-cutting realities: fold-down integrity, the object budget, loudness, remix-vs-re-render.
Genre rewrites the brief — captured versus constructed space — so each genre gets its own page.

The brief

A label wants an artist's new album released in immersive audio for the streaming platforms (Apple Music, Tidal), plus older singles from the back-catalogue reworked to match. Deliverables: a Dolby Atmos master (ADM BWF), a binaural render for headphones, and a stereo fold-down that must not sound worse than the original stereo. The mix happens in a small, treated studio on a 7.1.4 monitor layout. Budget is a few weeks, not a film schedule.

This is the studio end of the spectrum: no audience in the room, but every downstream listening situation — a phone, a soundbar, a car, a pair of earbuds — is out of your control. And it is not one job: the aesthetic contract differs so sharply by genre that each gets its own page. This hub covers what they share — the creative vocabulary and the delivery constraints — and hands off to the genre chapters for the rest.

The vocabulary of spatial creation — and its technical handles

Before any genre, there is a vocabulary. Working spatially with music means composing along a small set of perceptual axes, and each one has a concrete technical handle in this guide. Learning the pairing — the creative idea and the mechanism that realises it — is what stops immersive mixing from becoming knob-twiddling. This table is the spine of every genre page that follows; the artistic vocabulary develops the intention side, the linked chapters the mechanism.

Position — where a sound sits

The most basic decision: a location in the horizontal and (now) the vertical. Creatively it is figure-placement — the lead is here, the guitar there. Technically it is a panning law: amplitude panning (VBAP/DBAP) distributes a source's energy across the nearest speakers, and the ear fuses the result into a phantom image via inter-channel level and time differences (ICLD/ICTD). The creative caveat — a phantom image is only stable near the sweet spot — is a fact about the mechanism, and it is why placement that must survive off-centre (or on a phone) tends back toward the real speakers.

Distance and depth — how far away, how present

Depth is the axis amateurs ignore and professionals live in. A sound is pushed back not by lowering a fader alone but by three linked cues: falling level (roughly inverse-distance), a high-frequency roll-off from air absorption, and — decisively — a falling direct-to-reverberant ratio, more reverberant energy relative to direct (direct, diffuse & envelopment). Change these together and a source recedes convincingly; change level alone and it just gets quieter. Physically correct near-field depth (a source that feels in front of the speakers) additionally needs wavefront curvature, the domain of WFS.

Size and width — a point, or a body

A source is not always a dot. "How big is this sound" — a pinpoint hi-hat versus a pad that fills the room — is an expressive parameter, and its technical name is decorrelation. Feed near-identical signals to several speakers and the ear hears one narrow image; decorrelate them (independent phase/time relationships) and the image widens into an extended body. The perceptual measure of that width, Apparent Source Width, is governed by the interaural cross-correlation (IACC) of the sound reaching the two ears — low correlation, wide source. Practically this shows up as an object spread/size parameter, or RIPL's continuous Spread/Focus axis. The pitfall is also a fact about the mechanism: decorrelation that drifts toward anti-correlation collapses on mono fold-down.

Trajectory and motion — a path through time

Movement is a phrase, not a garnish (see gesture and trajectory). Technically a trajectory is position automation over time, and three mechanisms decide whether it sounds good: the interpolation of the panning gains and delays between keyframes (too coarse and you hear zipper noise; the renderer's interpolation order is a real quality lever); Doppler shift if the motion is physically modelled through a delay line; and the precedence effect, which for a fast move means the ear locks onto onsets, so rapid trajectories read as direction more than as precise position. Perception also sets a floor — below a minimum audible movement, automation satisfies the lane but not the ear.

Envelopment and the room — being inside a space

Distinct from placing sources is manufacturing the space they sit in. Envelopment — the sense of being surrounded and inside — is built from diffuse, decorrelated late energy, especially arriving laterally and from height, i.e. late reverberation spread around and above (direct, diffuse & envelopment). Crucially, room character is independent of source position: you can place a voice close and dry yet wrap it in a cathedral, which no natural space offers. The height layer, in most music, is this — air and hall — not a shelf for instruments.

Density — how much a listener can follow

The last axis is a budget. There is a hard perceptual limit on how many simultaneous, independently-placed spatial streams a listener can parse before they fuse into a mass (the density budget). Immersive tempts you to place everything; masking and this limit mean that past a handful of active spatial voices, more placement buys nothing. Deciding whether the listener should locate discrete sources or bathe in a field is a per-section choice.

The pairing, in one line

Position ↔ panning law · Depth ↔ direct/reverberant + air + level · Size ↔ decorrelation (IACC/ASW) · Trajectory ↔ automation + interpolation + Doppler + precedence · Envelopment ↔ diffuse late reverberation · Density ↔ the perceptual limit on simultaneous streams.

Genre rewrites the brief — a page each

"An immersive music mix" is really several jobs, separated by whether space is captured or constructed, and by where the voice may live. Each has its own chapter:

Classical & Orchestral — space is captured: fidelity to a real hall, perspective and envelopment from a microphone array, height as air. Objects barely used.
Jazz & Acoustic Small Ensemble — the ensemble in a plausible room: capture plus gentle placement, intimacy and depth over spectacle.
Pop, R&B & Hip-Hop — space is constructed: an anchored lead vocal, width by decorrelation, effects thrown into height, and the fold-down as the real master.
Rock & Band Music — the band as a believable stage: energy first, width and depth serving the song, live-capture realities.
Electronic & Experimental — space is the instrument: the full trajectory-and-size vocabulary, synthesis and spatialisation composed together.

Cross-cutting technical constraints

Whatever the genre, the same delivery realities bite:

The room is not the listener's room. A 7.1.4 studio is a reference; most listeners are on earbuds. Judge through the renders, and trust the binaural monitor path.
Fold-down integrity. Phantom-centre build-up, height collapsing into comb filtering, and the loudness relationship between immersive and stereo all surface only on the stereo and binaural fold-downs — check them, do not assume them.
The object budget and the bed. The Atmos music renderer works to a fixed budget of audio elements; deciding what is a discrete object (must localise or move) and what belongs in the bed (room, glue) is a real authoring choice.
Whose binaural renderer? The headphone result is not one thing — different platforms voice the same ADM differently. Check on the actual target chain.
Loudness and metadata. Streaming immersive normalises differently from stereo; reconcile the immersive and stereo levels, and treat ADM metadata as a deliverable (see Post-Production, Immersive Music Production).
Remix or re-render — and is it even deliverable? A true remix needs multitracks; with only a stereo master you are doing a principled re-render — a stereo decode, not channel invention. But mind the platform rules: Apple requires a Dolby Atmos music track to be built from multitracks or stems and prohibits upmixing a stereo mix as an Atmos deliverable. A stereo decode is the right tool where you control playback (live, installation) or as a mixing aid — not a shortcut around the stems requirement.
Capture decides your freedom. How the material was recorded sets how freely you can space it later: a captured field fixes the perspective, isolated sources let you place and move them. That upstream choice is a chapter of its own — see Capturing for Immersive.

Solutions — with RIPL

RIPL's relevance across every genre is its one unified source model: mono objects, native multichannel stems, and a decoded stereo field are all first-class sources in one scene, rendered by one engine to any output.

The vocabulary, as controls. The axes above map onto RIPL parameters directly: position and gain-and-delay placement, a continuous Spread/Focus for size, automation on the timeline for trajectory, and per-object interpolation that spends CPU on the moving foreground.
Captured or constructed, one tool. The classical record's captured perspective (a wide, distance-true stage plus a hall bed) and the pop record's constructed space (anchored vocal, placed and moving objects) are two uses of the same engine, not two products.
Catalogue reworks. Stereo masters enter through HSR, which separates localisable primaries from diffuse ambience before placement — so the vocal stays put and the ambience opens, with energy, coherence and timbre preserved.
One scene, every deliverable. Because authoring is separated from rendering, the Atmos feeds, the ambisonic archive, the binaural stream and the stereo fold-down are modes of one scene, not four mixes to keep in sync.

Pitfalls and checklist

Checklist

Did you pick the genre contract — captured fidelity vs constructed space — before placing anything?
Are you working the depth axis (level + air + direct/reverberant together), or only panning left–right and up?
Does any decorrelation for width survive mono fold-down, or does it collapse?
Is every trajectory above the audible-movement floor and free of zipper noise?
Is the density under what a listener can follow, or are you placing for placement's sake?
Did you verify binaural and stereo fold-downs on real devices and the platform's own renderer, and reconcile immersive vs stereo loudness?

Bibliography

Roginska, Agnieszka, and Paul Geluso, eds. Immersive Sound: The Art and Science of Binaural and Multi-Channel Audio. New York: Routledge, 2017.
Rumsey, Francis. Spatial Audio. Oxford: Focal Press, 2001.
Blauert, Jens. Spatial Hearing: The Psychophysics of Human Sound Localization. Rev. ed. Cambridge, MA: MIT Press, 1997.
Pulkki, Ville. "Virtual Sound Source Positioning Using Vector Base Amplitude Panning." Journal of the Audio Engineering Society 45, no. 6 (1997): 456–466.
Kendall, Gary S. "The Decorrelation of Audio Signals and Its Impact on Spatial Imagery." Computer Music Journal 19, no. 4 (1995): 71–87.

A note on sourcing: the reference list above gives well-established works; a few edition years are worth confirming against a primary source before print. Each genre page carries its own, more specific bibliography.

The brief​

The vocabulary of spatial creation — and its technical handles​

Position — where a sound sits​

Distance and depth — how far away, how present​

Size and width — a point, or a body​

Trajectory and motion — a path through time​

Envelopment and the room — being inside a space​

Density — how much a listener can follow​

Genre rewrites the brief — a page each​

Cross-cutting technical constraints​

Solutions — with RIPL​

Pitfalls and checklist​

Bibliography​

See also​