ASE

THE PROBLEM

Rays are blind below Schroeder

System design decisions are still mostly made on plans, experience and rules of thumb — then verified after the rig is hung, when every change costs real money and show time. The stakes of a design (coverage holes, intelligibility, low-end behavior) only become visible when it is too late to change them cheaply.

The simulation tools that do exist are blind exactly where rooms behave worst: geometric acoustics is fast but cannot see below the Schroeder frequency — room modes, low-frequency interference, the bass that defines small and mid-size venues. Wave methods see everything but were historically too slow to be practical.

What production workflows need is a decision tool that runs before the show: fast enough to fit inside normal prep time, precise enough to be trusted, and complete enough — real manufacturer speaker data, standard metrics — that the answer it gives is the answer the room will give.

HOW IT WORKS

How it works

UNDER THE HOOD

Going deeper

ASE runs every method where it is valid. The Image Source Method resolves first-order reflections exactly; adaptive stochastic ray-tracing carries the higher orders and the late field, multithreaded across cores; an FDTD solver simulates the wave equation itself below the Schroeder frequency, capturing the modes and low-frequency interference rays cannot see — with a smooth crossover joining the domains. Five quality levels scale the same physics from instant preview to reference-grade computation.

What separates ASE from the established simulators is phase. Most geometric engines sum energy; ASE's summation is coherent — complex pressure, phase tracked along every path, with a decoherence model that accounts for wind and temperature turbulence on long throws. Interference between sources, comb filtering, alignment errors: the effects that actually ruin systems appear in the simulation, exactly where they will appear in the room.

The physics goes to the edges of the model, literally: hybrid BTM + UTD diffraction handles multiple edges with frequency-dependent coefficients, keeping shadow zones and balcony overhangs plausible where simpler engines go silent. And the air itself is modeled to the standard: full ISO 9613-1 atmospheric absorption — temperature from −20 to +50 °C, humidity, pressure, altitude, the oxygen and nitrogen relaxation that shapes every long throw above 2 kHz.

Every computed quantity renders as a layer over the venue: direct SPL coverage, time alignment and arrival delays, reflection patterns, room modes with their Q factors, and the most complete metric set in its class — RT60 (Sabine, Eyring, Fitzroy and measured from the IR), EDT, C50/C80, D50/D80, centre time, lateral fractions Lf/Lj, stage support ST1/ST2, ITDG and echo criterion, plus STI, RASTI, %ALcons and 18-band SII intelligibility to IEC 60268-16 and ANSI S3.5. You check the design against the standard and the spec, seat by seat, before anything is hung.

Reverberation is predicted with engineering precision: the ray-traced tail and the FDTD modal behavior combine into per-seat impulse responses, and coupled-volume models reproduce the double-slope decays of foyers and naves. Sources are real loudspeakers — manufacturer GLL / CLF data, from L-Acoustics and d&b to community-published files, with third-octave directivity balloons at 5° resolution, multi-way sub-sources and transfer functions — so the simulation runs on the system you will actually hang.

And the data does not stop at diagnosis: from the computed coverage and timing, the system can be calibrated automatically — speaker placement, gains, delays and per-output FIR equalization derived from the simulation rather than tuned by trial and error on site. Setup is fast, computation is faster than real time: between the two and the cost of the application, every production — not just the flagship ones — gets a result validated to an exemplary standard before load-in.

THE SOLVERS

Three solvers, one physics

No single method covers a room from 20 Hz to 20 kHz. ASE assigns each frequency range to the solver that is physically right for it, and joins them with a phase-aligned crossover whose split point follows the room itself — around 500 Hz for a small control room, 100–150 Hz for a large hall, tracking the Schroeder frequency.

PHASE

The simulation hears interference

Most geometric simulators add energies. Real systems add pressures — and the difference is everything that goes wrong at soundcheck. ASE sums coherently: complex pressure, phase tracked along every path from every source, so interference between boxes, comb filtering and alignment errors appear in the prediction exactly where they will appear in the venue.

Coherence is also not pretended where physics destroys it: a decoherence model accounts for the travel-time fluctuations that wind and temperature gradients impose on long throws, so distant interference is weighted by how stable it would actually be. Energy-only summation remains available where it is the right model — the point is choosing, not guessing.

AT THE EDGES

Diffraction and air, modeled to the standard

METRICS

Everything the standard can ask of a room

Computed per band, at every receiver, from per-seat impulse responses — and rendered as layers over the venue.

GEOMETRY & MATERIALS

Import the venue as it is

The room comes in as a standard 3D model — OBJ or GLTF, straight from the venue's drawings or a quick model. Visual materials map automatically to acoustic ones: the surface named concrete in the file becomes concrete in the simulation, with per-band absorption and transmission you can override surface by surface.

Transmission matters as much as absorption: sound that passes through a stage curtain or a lightweight wall is part of the answer, and ASE carries it through occlusion analysis instead of treating every surface as a perfect barrier.

CALIBRATION

From prediction to preset

Because the model is phase-accurate and per-seat, it contains everything a calibration needs. ASE derives speaker placement, gains, delays and per-output FIR equalization directly from the computed field — the same physics that predicted the problem produces the setting that fixes it. The result loads as a starting point that is already coherent, before a single measurement microphone comes out of its case.

PERFORMANCE

Reference quality at preview speed

Five quality levels scale the same physics from instant preview to reference-grade runs — move a box and watch the coverage follow, then let the full computation validate the final design. Everything is multithreaded C++17, faster than real time on a laptop; on the GPU roadmap (Metal and CUDA), the FDTD stage of a 20 × 15 × 5 m hall drops from minutes on CPU to seconds.

AT A GLANCE

At a glance

PARAMETERS

Hands on the algorithm

POSITIONING

Compared to the alternatives

APPLICATIONS

Where it fits