Aller au contenu principal

Case Study — Live Concert Spatialisation

In brief
  • A touring amplified show for a standing, moving crowd — no sweet spot, and the rig changes nightly.
  • Use wide-area placement (DBAP/WFS), not sweet-spot phantom imaging.
  • Anchor the visible band frontally; compose a few legible signature moments.
  • Author once as objects, render to tonight's array, drive from timecode; keep latency low and constant.

The brief​

A band with electronic and acoustic elements tours mid-size venues (1,500–3,000 capacity, standing). They want more than left–right: instruments placed across a widened stage image, a few signature spatial moments (a synth that circles the room, a delay that throws to the back wall, a backing vocal that lifts overhead), and — crucially — it must read as intentional from the barrier to the back bar, not just at the mix position. The rig changes shape every night: a clean hang in one room, a compromised one under a balcony in the next, a festival slot with whatever is already flown. The show is cue-heavy and timecode-locked to lighting and video.

This is the opposite pole from the studio: a real audience, spread across the whole room, none of them in the sweet spot, in a space you did not design and only meet at load-in. Everything below follows from that one fact.

Creative problematics and reflections​

  • Decide the spatial identity of the show before the technology. The first question is dramaturgical, not technical: which sources are placed out in the room, and which stay in the powerful frontal PA that anchors the band? A spatialised concert is emphatically not "everything moving." It is a clear frontal anchor for the visible musicians plus a small, deliberate vocabulary of placed and moving elements. This is the sweet-spot-versus-coverage decision from diffusion logics resolved firmly on the coverage side, and it is the same standing-audience problem treated at length in Electronic & Live Music — read that chapter first for the artistic framing this case turns into a workflow.

  • Compose moments, not wallpaper. A standing audience reads a few strong spatial gestures — a slow sweep, a sudden widening, a voice arriving from behind — far more reliably than continuous motion, which simply averages out into an unlocalised wash. Choose the three or four moments in the set that mean something and make each unmistakable. The discipline here is exactly the "gesture as phrase" idea from Thinking the Sound Space: a spatial move should have a shape, a direction, and an arrival, not merely happen.

  • The visible band is your anchor — use it. Because the audience can see the players, vision anchors the frontal image for you (the audio-visual coherence discussed in Stage & Live Performance). That is what licenses the spatial play: a stable, seen frontal source lets one clear gesture stand out against it. Detach everything and the crowd has nothing to hold onto; anchor the band and the travellers become legible.

  • The crowd is the instrument's room, and it is hostile to subtlety. Bodies absorb high frequencies, the PA is loud, reflections are everywhere, and no one is standing still. Gestures that would be exquisite in a listening room die here. Whatever you design, assume it must survive being heard off-axis, at a distance, through a wall of people — which means bigger, slower, and more contrasted than studio instinct suggests.

  • Restraint reads as confidence. The failure mode of live immersion is the channel-count demo — sound flung around because the rig can. Every placed or moving element should answer to the music, not to the spec sheet. If a move would be invisible to a listener at the back bar, it is cost without benefit; cut it.

The cases within the case: the genre sets the rig​

  • Rock & pop band. A strong frontal L/R (or L/C/R) PA carries the power and the vocal; spatialisation is a layer — widen keys and guitars, place backing vocals, throw a few effects — over a mix that must still hit hard mono-compatibly. The lead vocal almost always stays anchored front-centre for intelligibility.
  • Electronic act / DJ. The most spatial case: no acoustic origin, often no visible instruments, so the whole field is fair game and the audience expects immersion. Content can be authored as objects in advance and much of the "performance" is spatial. But energy and low-end impact still rule — a clever pan that kills the drop is a failure.
  • Amplified / hybrid orchestra & acoustic acts. The goal is natural placement — the string section where it sits, reinforced discreetly so the image matches the stage picture — closer to theatre's source-oriented reinforcement than to effect-driven motion.
  • Singer-songwriter / small acoustic. Minimal, tasteful width and depth; the risk is gimmickry. Often the spatial win is simply an even, enveloping reverb return rather than moving objects.

Technical problematics and reflections​

  • There is no sweet spot, so phantom imaging is the wrong primary tool. A stereo pair images beautifully at the desk and collapses to the nearer speaker everywhere else — the "everyone but the centre gets mono" problem at the heart of live sound. With a standing crowd covering the whole floor, the system must place sources robustly across a wide area, which pushes you away from inter-channel phantoms and toward wide-area panning and wavefront methods (see amplitude panning and WFS).

  • Coverage, delay and the precedence effect decide everything. An immersive PA is many arrays plus delays, and whether an image holds or smears is governed by time alignment and phase and by the precedence (Haas) effect from psychoacoustics: the ear localises to the first wavefront, so getting the relative timing of mains, out-fills, and delays right is what pulls the image to the stage rather than to the box nearest each listener. Speaker plot and coverage mapping — the material of speaker layouts & topologies — are the foundation the spatial layer sits on; no renderer rescues a bad hang.

  • Latency must be low and constant. A live band feels its own sound. The path through the spatial processor has to add minimal, stable delay — jitter in latency is worse than a fixed offset, because it moves the image and unsettles the players. This constrains the choice of rendering method and the network design.

  • The low end is not spatial, and must not pretend to be. On a big system sub-bass is felt in the body and localises poorly; spatial interest lives in the mids and highs over a bass that is essentially mono and everywhere. Treat the subs as their own coverage problem — see subwoofers & bass management — and put your spatial gestures above them, not in them.

  • Robustness over precision, because the room changes nightly. A trajectory authored for one geometry must degrade gracefully when tonight's room is wider, the surrounds are sparser, or a delay position is missing. The design has to hold its intent under re-rendering, which is only possible if the show is described abstractly rather than patched channel-by-channel.

  • Show control is a hard requirement. Spatial states must recall on cue, in sync with lighting and video timecode, and survive a festival changeover onto a different array count. A spatial design that cannot be stored, recalled, and handed to tomorrow's system is not tourable.

Solutions — the general approach​

Use a wide-area placement method rather than sweet-spot phantom imaging: amplitude-panning families such as DBAP that spread a source's energy so localisation holds across the whole floor, and WFS on dense front arrays where you want genuine depth and sources that seem to stand in front of the PA. Choose the method per zone and per source, not one paradigm imposed on everything — the general principle set out in diffusion logics.

Author the show as an object scene with cued spatial states so the same design re-renders to each night's array, and drive those states from timecode alongside lighting and video, exactly the live-sound workflow extended into space. Keep the frontal reinforcement — the band's anchor — structurally primary and time-aligned first (time alignment & phase); layer the placed and moving elements on top of a system that already covers every seat. Give near-field, in-the-crowd sources a sense of distance with the cues from distance & air rather than level alone.

This object-plus-renderer model is shared across the commercial immersive-PA stacks — L-ISA, d&b Soundscape, Holophonix, and DAM Audio's RIPL — so the workflow and the thinking transfer whichever platform a given tour carries.

Solutions — with RIPL​

  • One design, every venue. The show is authored once as a RIPL object scene. Because authoring is separated from rendering, RIPL decodes that one scene to tonight's actual array — a dense front line rendered WFS-style for real depth, sparser surrounds filled with robust gain-panning — with no re-authoring at load-in. The abstract description is the tour asset; the render is disposable and rebuilt per room.

  • Gain and delay as two controls of one source. RIPL treats amplitude panning and wavefront delay as two knobs on a single unified source model, so you are not forced to choose one paradigm for the whole show. A near-field synth can be given genuine depth via delay, while a wide frontal source is given crisp placement via level — the right blend of time and level per source, rather than a single method's limitations imposed everywhere.

  • Cued, timecoded, and live at once. Spatial positions live on the timeline as automatable parameters: the circling synth and the thrown delay are drawn once, stored as cues, and recalled to timecode next to lighting and video — or nudged live from a tablet, the modern descendant of the acousmatic diffuser's desk (see Electroacoustic & Acousmatic Music for that lineage). The same move can be composed in advance, performed live, or both.

  • CPU spent where the ear is. RIPL prioritises its expensive interpolation on the loud foreground object the crowd is actually tracking and economises on the background, keeping fast moves artefact-free without overloading a live engine that has no second chances.

  • A redundant AoIP backbone. The scene rides a standard audio-over-IP network with redundancy — which is what makes a fast festival patch and a fail-safe show survivable, and what lets the same design travel from a clean club hang to a compromised shed without rebuilding the signal chain by hand.

Pitfalls and checklist​

Checklist
  • Did you walk the room? A gesture that only reads at the desk is not a gesture for a standing crowd. Listen from the barrier, the sides, the back bar, and under the balcony before you trust any spatial move.
  • Is there a clear frontal anchor, or is everything floating? If the band is not solidly imaged to the stage, the placed elements have nothing to stand out against.
  • Is processor latency low and, above all, constant enough for the band's feel and monitors?
  • Do your cues recall correctly on a different array count at a festival slot, or does the design assume last night's rig?
  • Are the signature moments big enough to survive a loud, absorptive, moving crowd — or are they studio-scale subtleties that will vanish?
  • Is the low end left mono and coverage-managed, with the spatial play kept above it?

See also​

In the technical guide

In artistic practice

→ Next: Theatre & Live Performance