Checked against a measured roomAll effects

Physical Room Reverb · Every Room Solved

Hangar

Research Grade Reverb.

Give it a room, its shape, its surfaces and the air, and Hangar works out how sound travels through it: image sources for the early reflections, rays for the late energy, and the wave equation for the low octaves. Then it shows you every path it found, and how its answer compares with a room somebody measured.

VST®3 effect · Windows 64-bit No activationmacOS VST3 and Audio Unit are in development.

The ROOM page of the Hangar panel in dark glass: the STONE CHURCH from the library drawn in orbit, its barrel vault, its columns and its pews, a singing voice at the front and an ORTF pair in the nave, with the loudest reflection paths and the panel's own rays drawn between them; on the right the ROOM card with the church's dimensions and the air, the OBJECTS card listing its columns and pews, and WHY IT SOUNDS THIS WAY with the church's account.
  • 41 rooms
  • 75 materials
  • 4 sources
  • 0 latency
The ROOM page of the Hangar panel with the STONE CHURCH, which the camera moves across part by part: the top bar with the solve, the latency and the MIX light, the view with the reflection paths and rays, and the ROOM, OBJECTS and WHY IT SOUNDS THIS WAY cards.

THE ROOM PAGE

A stone church, solved.

A library room as it opens: a nave under a barrel vault, an aisle each side behind its columns, the pews, a voice at the front and a pair of microphones in the nave. Every surface its material, every path the sound takes drawn in the room.

Three solvers, one room

Solved, not imitated.

An algorithmic reverb imitates a room's decay; a convolution reverb plays back a room somebody once recorded. Hangar does neither. It takes the room you give it, its shape, its surfaces, the air, the sources and the microphones, and works out how sound travels through it, each part of the response by the method that is right for it.

THE FIRST TENTH OF A SECOND

IMAGE SOURCES

Every early reflection found exactly, by mirroring the source in the walls: Allen and Berkley's lattice in a box, Borish's tree in any other shape, each image checked for what it can see. Each arrives at its own time, from its own direction, with each surface's absorption by octave and the air's along the way.

THE LATE ENERGY

RAYS

Rays leave the source as its directivity sends them and reflect as each surface's scattering says, mirror-like or spread, losing what each surface and the air take. Spheres at the microphones collect them by time, octave and direction, and the late tail is drawn from what they collected.

BELOW THE CROSSOVER

THE WAVE EQUATION

Where a room is a handful of resonances, a wave solver computes the sound field itself, every material's absorption fitted as a passive admittance on the grid, and the walls rough where their materials scatter. The low end comes out with its modes, its interference and its diffraction, because they are in the equation.

A third-order Butterworth crossover joins the wave solver to the geometry, its two halves adding to one in level and in power, so the late field is right at the crossover as well as the early sound. One calibration serves every room, so a cupboard and a cathedral compare honestly.

Why the low end needs waves

Below a frequency, a room is notes.

High up, a room's resonances overlap so densely that its response is a statistic, and rays describe it well. Low down they stand apart, each its own note, and only the wave equation gets them right. The line between is the Schroeder frequency. Shape the room and see where it falls.

EVERY MODE OF THE BOX, SUMMEDAXIALTANGENTIALOBLIQUE
2050100200500TOO DENSE TO DRAWSCHROEDER4 × SCHROEDER

79 MODES BELOW 149 HZ, THE SCHROEDER FREQUENCY OF A ROOM OF 144 M³ DECAYING IN 0.80 S.

A drawing of the arithmetic, computed here: a rigid box's modes, f = (c/2)√((nx/W)² + (ny/L)² + (nz/H)²), each drawn as a resonance 2.2/T60 Hz wide and every one excited alike, and the Schroeder frequency, 2000√(T60/V). Hangar's crossover follows the same rule, four times the Schroeder frequency and at least 200 Hz, as far as the wave solver's budget on the machine allows; the shaded part of the drawing is where its wave solver would carry the sound. Its own checks hold the wave solver's modes in a rigid box to this formula within 1 percent.

Rough walls

The low octaves diffuse too.

A wave solver with perfectly smooth walls makes a small room ring on in its low octaves far longer than the rays, which scatter by each surface's coefficient, say it should. So where a material scatters, Hangar draws its walls rough on the wave solver's grid: blocks standing proud and recesses cut in, fitted to the material's scattering coefficient. Their absorption stays the material's own.

0.250.500.751.00631252505001K2KTHE COVERAGE: WHERE IT SATURATES

SCATTERS 0.11 AT 125 HZ, 0.29 AT 250 AND 0.44 AT 500, AND NEVER MORE THAN ITS COVERAGE, 0.50.

A drawing of the arithmetic, computed here: Kirchhoff's specular loss for a rough surface, averaged over the angles sound arrives from and taken on the covered part only, s = φ(1 − (1 − e−g)/g), g = 4k²σ², with σ, the blocks' RMS height, 1.04 times the depth. Hangar fits each material's coverage and depth to its coefficient in the octaves the wave solver carries, no deeper than 0.4 m or a tenth of the room's smallest dimension. In Hangar's own test rooms the relief brings the wave solver's decay close to the rays' at 250 and 500 Hz. At 125 Hz a relief as deep as a room allows scatters less than the rays assume, and there the decay can still run long.

Checked against a measured room

A real room, and our answer.

BRAS, the Benchmark for Room Acoustical Simulation, publishes real rooms with their measured responses and the surfaces fitted to them, so that simulations can be compared with the truth. We solved its seminar room, CR2, and analyzed the measurements and our solve with the same code.

MEASURED T30HANGAR'S SOLVETHE CLAIM, 500 HZ TO 4 KHZ: WITHIN 10 PERCENT
0.5 S1.0 S1.5 S2.0 S125 HZ+28.9%250 HZ-4.7%500 HZ-2.0%1 KHZ+5.7%2 KHZ+4.6%4 KHZ+6.4%
BRAS CR2: measured and solved, the mean of ten source and microphone pairs, by octave
OCTAVET30 MEASUREDSOLVEDEDT MEASUREDSOLVEDC80 MEASUREDSOLVED
125 HZ1.42 S1.83 S +28.9%1.39 S1.78 S +28.1%+1.6 DB-1.4 DB -3.0NOT CLAIMED
250 HZ1.72 S1.64 S -4.7%1.46 S1.69 S +15.8%+1.1 DB-0.6 DB -1.7NOT CLAIMED
500 HZ2.02 S1.98 S -2.0%1.99 S2.03 S +2.0%-1.3 DB-1.2 DB +0.1CLAIMED
1 KHZ1.94 S2.05 S +5.7%1.90 S2.00 S +5.3%-0.8 DB-0.9 DB -0.1CLAIMED
2 KHZ1.74 S1.82 S +4.6%1.72 S1.84 S +7.0%-0.1 DB-0.3 DB -0.2CLAIMED
4 KHZ1.57 S1.67 S +6.4%1.58 S1.67 S +5.7%0.0 DB0.0 DB 0.0CLAIMED

THE CLAIM

From 500 Hz to 4 kHz, Hangar's T30 and EDT are within 10 percent of the measurements and its C80 within 1 dB, octave by octave, as the mean of the ten source and microphone pairs BRAS measured with its dodecahedron. Both were analyzed by the same ISO 3382-1 code. Hangar's own checks run this comparison again whenever the BRAS data is beside them.

ITS SCOPE

BRAS gives CR2's geometry only as model files, so Hangar solved it as an equivalent box: CR2's floor area and volume, its fitted floor and ceiling, and the walls carrying the absorption of its plaster, concrete and windows. At 125 Hz the box still rings longer than the room, and at 250 Hz its early decay runs long: the real room diffuses through geometry the box does not have, its deep window reveals, a column and a recess. Those octaves are not claimed. BRAS's authors note that its complex rooms carry more measurement uncertainty than its reference scenes, and are best used to compare simulations given the same input.

L. Aspöck, F. Brinkmann, D. Ackermann, S. Weinzierl and M. Vorländer, BRAS: Benchmark for Room Acoustical Simulation, RWTH Aachen University and TU Berlin, 2020, published under CC BY-SA 4.0. CR2 is a seminar room at RWTH Aachen University. The data was used for this comparison and is not part of Hangar.

The library

Forty-one rooms, each explained.

A preset in Hangar is a room: its geometry, its materials, its objects, the air and the microphones, never a recording. Every room in the library says why it sounds the way it does, and any of them is a starting point: change a wall and the room solves again.

STUDIOS

  • VOCAL BOOTH

    Fabric on every wall, so almost nothing comes back; the booth's few high modes make a voice close and a little boxy.

  • LIVE ROOM

    Hard walls facing soft ones, so no two parallel surfaces flutter, under an acoustic tile ceiling.

  • WOOD STUDIO

    Thin wood over air takes the low mids, warm rather than boomy, and a diffuser breaks up the back wall's slap.

  • DRUM ROOM

    Small and hard: the first reflections land within moments of the hit and stack on it.

  • STONE ROOM

    Stone barely absorbs, so the decay is long for the size and nearly one color from the bass to the treble.

  • PLASTER CHAMBER

    A small plaster room built only to ring: about 4 seconds at middle frequencies, its bass ringing on the room's own notes.

  • CHURCH STUDIO

    A church turned studio, plaster and bare boards under a high ceiling: about 8 seconds, the board floor taking a seventh off it.

DOMESTIC

  • CUPBOARD

    A rail of coats takes the top; what is left is a handful of modes, the honk of shouting into a cupboard.

  • BATHROOM

    Tile reflects nearly everything, and its few strong low modes are the notes a voice in the shower finds.

  • LIVING ROOM

    Soft furnishings take the treble, and the sofa and the bookcase break up the reflections, so it never flutters.

  • CAR INTERIOR

    Glass all round and upholstered seats, with a binaural head in the driver's seat.

  • BRICK CELLAR

    A barrel vault focuses: under the crown the return is stronger and more colored than at the side.

  • STONE STAIRWELL

    A drum kit at the foot of a stone stair hall: from the first floor up, the stair cuts off its direct sound, leaving only the room.

PUBLIC AND INDUSTRIAL

  • STAIRWELL

    The flights and landings scatter, and the shaft's height gives a long, rising decay.

  • PARKING GARAGE

    Sound spreads between two hard plates and the columns scatter it: a long, grainy tail with no walls in it.

  • GYMNASIUM

    Hard floor, walls and roof hold a note for over 7 seconds; the bleachers along one side keep the long walls from fluttering.

  • WAREHOUSE

    Thin steel flexes, taking the low end and reflecting the top: long, bright and a little hollow.

  • TUNNEL

    Brick and concrete, its open ends drawn as wool: the bass lasts over 4 seconds and the middle under 2.5.

  • SILO

    A round wall focuses sound back to the middle, and off the axis the echo creeps along the wall.

  • WATER TANK

    Water reflects almost everything, and a round tank under a dome focuses as well as rings.

  • OIL TANK

    A sealed concrete tank with nothing soft in it: the bass takes about two minutes to die away, the middle about half a minute.

  • COLUMN CISTERN

    Water, brick and stone barely absorb, so the decay runs past 10 seconds, and the columns block most long paths: no clear echo.

  • TERMINAL CONCOURSE

    Stone, marble and a hung plaster vault: about 7 seconds with a standing crowd on the floor, and about 14 without it.

  • GREAT WAITING ROOM

    Stone and coffered plaster around 130,000 cubic meters: about 14 seconds, the travelers on its benches the main absorber.

HALLS

  • CHAMBER HALL

    A narrow shoebox: strong early reflections from the side walls, and the sense of being surrounded.

  • CONCERT HALL

    The audience is the hall's main absorber, under a wooden organ wall, with a Decca tree above the conductor.

  • AUDITORIUM

    Splayed walls send the side reflections to the back: clear for speech, and less enveloping.

  • GILDED SHOEBOX

    Gilded relief scatters what smooth walls let run, so the sound keeps reaching the audience and the audience sets the decay.

  • COFFERED SHOEBOX

    Coffers and niches scatter everything above a few hundred hertz down into the audience, which does nearly all the absorbing.

  • WIDE SHOEBOX

    Half again as wide as the gilded shoebox, so its first side reflections reach the microphones some 25 milliseconds later.

  • ELLIPTICAL HALL

    A dome focuses the sound back down an eighth of a second late, a distinct echo; sixteen hung discs catch and scatter most of it.

SACRED

  • STONE CHURCH

    The columns scatter, the aisles return their own later energy, and the stone barely absorbs.

  • VAULTED CHAPEL

    Plaster, stone and glass barely absorb, so a chord hangs for 5 seconds even with wool tapestries taking the middle and the treble.

  • FAN VAULT CHAPEL

    Stone and leaded glass ring for about 7 seconds; beyond the organ screen, the choir arrives by way of the vault, quieter.

  • CATHEDRAL

    Its volume is the whole story: the air takes the treble long before the walls do.

  • MINSTER NAVE

    A long Gothic nave under a timber vault: boards and leaded glass take the bass, and the middle decays in about 6.5 seconds.

  • ROTUNDA

    A dome is a concave mirror: from the center the sound comes back focused, and a whisper creeps round the wall.

  • OCULUS ROTUNDA

    Marble and a concrete dome, open at the crown: about 18 seconds at 1 kilohertz, with a third of the bass lost through the opening.

NATURAL

  • CAVE THEATRE

    Wet rock barely absorbs and its facets scatter, so the long decay comes back smooth; the passage to the mouth takes the bass.

THE HANGAR

  • THE HANGAR

    So long that the first reflections arrive as distinct echoes before the tail closes in; the flexing steel takes the bass.

  • AIRSHIP HANGAR

    THE HANGAR's steel and concrete at seventeen times the volume: quieter at first, but about 24 seconds at 1 kilohertz against 16.

Six views

Everything it found, on the panel.

Six pages, and a view of the room on every one of them, in three cameras and with four layers: the paths, the rays, the wave field and the coverage.

MATERIALS The material in hand, its absorption and scattering by octave, the library by kind, and the room's decay three ways: SABINE, EYRING and as solved.

And the rest of it

Built to be checked.

What a reverb usually keeps to itself, Hangar puts on the panel.

MOTION

A source that moves.

Move a source and its direct sound and early reflections are computed as it goes, each taken from where the source was when the sound left it, so its Doppler shift is exact. What it sent before keeps arriving through the room, so nothing is heard twice; when it stops, the room solves where it stopped.

MICROPHONES

Thirteen arrays.

MONO, XY, BLUMLEIN, MID-SIDE, ORTF, NOS and a SPACED PAIR; the DECCA TREE; 5.1, 7.1 and 7.1.4 IMMERSIVE; FIRST-ORDER AMBISONICS; and a BINAURAL head from the University of York's SADIE II database. Each capsule has its own pattern, from omni to figure-8, and the late field reaches each as its pattern hears it.

THE SHARED ROOM

One room for the whole mix.

Choose a group, A to H, and every Hangar in the session on it shares one room: one scene, every track's sources in it, one change heard on every track. Up to 32 instances on a machine, and the room solves once for the group.

ZERO LATENCY

Nothing to wait for.

The convolution starts on the first sample. Its long blocks are spread across the audio calls inside them rather than handed to threads, so the load stays smooth and an offline render plays the same numbers as playback. A new solve crossfades in, never a click.

THE GPU

The wave solver, on the card.

Where there is a graphics card the wave solver runs on it, through Direct3D 11, and gives the same answer as the CPU's, to within rounding. Without one, or if the card is lost mid-solve, the same room solves on the CPU. The solve comes in stages, so you hear the room at once and hear it sharpen.

OBJECTS

A plate and a spring.

A plate built from its own modes, its metal, size, thickness and damping yours to set, and up to three springs of dispersive chains you can KICK. Either stands alone, or in the room, its outputs playing as two of the room's sources.

THE MIX

Honest about what it changes.

The MIX page has what a mix needs and a room does not: the direct sound's level, pre-delay, the early and late balance, shelves, a high-pass, width and a transformer or tube color. It is marked NOT PHYSICAL, and the top bar's MIX light says IN USE the moment any of it leaves neutral, so nobody mistakes an adjusted room for a solved one.

MATERIALS

Seventy-five surfaces, sourced.

Seventy-five materials in twelve kinds, from rendered brick to an audience on wooden chairs, with their absorption by octave from ISO 354 measurements as compiled in Vorländer's Auralization; the few not in it come from other tables or are estimated, and are marked so. Take any of them, copy it and make it your own.

YOURS

No activation.

The purchase lives at the store; the plugin itself never asks who you are, has nothing to unlock and never phones home. Mono, stereo or quad in; mono to 7.1.4 or first-order Ambisonics out. VST3 on Windows now; the macOS VST3 and the Audio Unit are in development.

The physical room reverb

Hangar

Research Grade Reverb.

VST®3 effect · Windows 64-bit No activationmacOS VST3 and Audio Unit are in development.

At a glance

Hangar specifications

Technical specifications for Hangar, Physical Room Reverb · Every Room Solved
FormatVST®3 effect · Windows 64-bit
In developmentmacOS VST3 and Audio Unit
Requires64-bit Windows 11, in any version Microsoft still supports. When the Mac version ships: macOS 15 or later, on Apple silicon or Intel.
The solverImage sources, rays, and a wave solver below the crossover
Wave solverDirect3D 11 compute on the GPU, or the CPU
Rooms41, each with its account · 16 templates and a drawn plan
Materials75 · ISO 354 measurements, the estimates marked
SourcesUp to 4, with directivity · moving, with exact Doppler
Microphones13 arrays · mono to 7.1.4, first-order Ambisonics, binaural
InputsMono, stereo or quad
OutputsMono, stereo, quad, 5.0, 5.1, 7.0, 7.1, 7.1.4 or first-order Ambisonics
LatencyZero
AnalysisISO 3382-1 by octave, 63 Hz to 8 kHz · changes marked in just-noticeable differences
Measured againstBRAS seminar room CR2 · T30 and EDT within 10 percent, C80 within 1 dB, 500 Hz to 4 kHz
Shared roomGroups A to H · up to 32 instances
ObjectsA plate and a spring, alone or in the room
The panel1600 by 940 · ROOM, MATERIALS, STAGE, ANALYSIS, MIX and OBJECT
Validation47 of 47 validator tests · host-tested
ActivationNone. No license server, no phone-home
Security updatesUntil November 2031
LicenseLicense agreement · Facts sheet
VST Compatible

VST is a registered trademark of Steinberg Media Technologies GmbH.