Public beta — 2.12.0

Design the filter.
Then print the page
that goes in the rack.

Alpha Sound Lab takes a measurement, aligns a multi-way system, fits the correction, and works out the limiter settings each driver actually needs. Then it prints all of it on one page you can type into any processor — including every processor that has never heard of FIR.

Download the beta See the settings sheet

Windows 10 1809 or later · 30-day demo · no account, no telephone home

The workflow

Eight stages, in the order you actually work

Import a measurement. Align the ways. Set the crossovers. Fit the correction by hand or automatically. Listen to it. Export it. Each stage shows you its result before you move on — there is no button that does everything, on purpose.

1Import
Smaart, REW, or any file with frequency, magnitude, phase and coherence columns. Mic calibration is applied once, and the tool records who applied it.
2System
Sub, low, mid, high. Delay, polarity and gain per way, with the alignment worked out from where the ways actually overlap.
3Target
Flat, tilted, shelved — the curve you are correcting towards, owned in one place.
4FIR Design
Windowed-sinc and linear-phase Linkwitz-Riley legs, with the delay and polarity carried in the taps.
5IIR / PEQ
Parametric bands by hand, with Q and bandwidth in octaves side by side.
6Auto Fit
Levenberg-Marquardt fitting that prefers few bands to many, and refuses to correct what it cannot see.
7Audition
Hear the correction on your own material before it goes anywhere near a rig.
8Export
Coefficients, processor presets, and the settings sheet.

The part nobody else does

A page you can type into any processor

Every FIR designer assumes the job ends with a coefficient file loaded into a FIR-capable box. Most boxes in the world are not that, and the engineer standing in front of a DriveRack is not going to buy a rack to use a measurement tool.

So the tool prints the numbers instead: crossover points and slopes, delays in milliseconds and samples and metres, polarity, gain, the band list with Q and bandwidth both, and the limiter settings for every way. A system set up from that sheet and a DriveRack gets the alignment, the crossover and the EQ. It gives up only the linear phase.

Prints, emails, or survives being photographed on a phone
Readable six months later without this program installed
States what it does not model, so nobody treats it as a guarantee

================================================================
  SYSTEM SETTINGS SHEET
  Main hangs - 4-way line array
  48000 Hz    343 m/s assumed for distances
================================================================

  WAY 4   High 1.4" horn
----------------------------------------------------------------
  Crossover      high-pass  1.60 kHz  Linkwitz-Riley 24 dB/oct
                 (no low-pass - this way runs to the top)

  Delay          4.104 ms   197.0 samples   1.408 m / 4.62 ft
  Polarity       normal
  Gain           -4.5 dB

  RMS limiter    threshold  -1.72 dBu   -3.94 dBV   0.635 V
                 attack 0.62 ms      release 100 ms

  Working        driver 80 W AES / 8 ohm  ->  25.30 V
                 amp 700 W / 8 ohm, 32.0 dB gain
                 timing from one period of 1600 Hz
                 at the threshold the driver sees 80 W

Limiters

The same rig. Thirteen decibels apart.

An 18″ sub and a compression driver on one system want completely different limiter thresholds, and a single setting is wrong for both. The threshold is arithmetic — the driver’s rating referred back through the amplifier’s voltage gain — so the tool computes it per way rather than leaving you to guess.

18″ sub · 1600 W AES · 4 Ω

+8.28dBu

+6.06 dBV · 2.010 V
attack 33.3 ms, release 333 ms

1.4″ horn · 80 W AES · 8 Ω

−1.72dBu

−3.94 dBV · 0.635 V
attack 0.62 ms, release 100 ms

Give the horn the sub’s number and it sees

13.01dB too much

which is 10 × log₁₀(1600/80), exactly.

Three units, always, on the same line

Processors disagree about whether their threshold field means dBu, dBV or volts, and dBV sits exactly 2.2185 dB below dBu. Reading one into the other’s field is 2.2 dB hot — which a sub shrugs off and a compression driver does not. So every threshold is printed in all three and nobody converts anything at one in the morning.

And it refuses to guess

No driver rating, no suggestion. Thermal time constants, excursion below box tuning, and what a real amplifier does into a real load are named on every sheet as not modelled, because a threshold presented without them is a number people treat as a promise.

Several positions

What EQ can fix, and what it cannot

A measurement taken at one place describes one place. Part of what it shows is the loudspeaker, and EQ fixes that everywhere. The rest is the room, and EQ “fixes” that at the microphone while making every other seat worse. One measurement cannot tell them apart — not because it is a poor measurement, but because the information is not in it.

Measure the same source from several seats and the tool separates them. What is the same everywhere is the loudspeaker; what moves as you move is not.

Six seats, one loudspeaker — the example set that ships with the tool
Frequency	What it is	Spread across seats	Coherence	Verdict
94 Hz	port tuning	2.1 dB	0.97	EQ it
331 Hz	reflection null	20.8 dB	0.97	leave it
1804 Hz	horn dip	6.4 dB	0.97	EQ it

Coherence is 0.97 at all three. Coherence tells you how good the measurement is, not whether the thing it measured exists anywhere else — a 20 dB null in one seat can be a flawless measurement. Without the position check, a fitter would happily correct all three.

Editions

Buy it once

Perpetual licence. No subscription, no seat rental, no account.

Standard

One loudspeaker at a time

Measurement import and mic calibration
Target curves
FIR design and PEQ fitting
Audition
Coefficient and processor export

Everything

Suite

Multi-way systems and racks

Everything in Standard
Multi-way system tree with per-way delay and polarity
Alignment, and how exact it has to be
Per-way limiter and compressor settings
Several-position analysis
The settings sheet

Pricing is being set before general release. During the public beta both editions are open — see below.

Public beta

Try it for thirty days

The demo is the whole program. It imports, designs, aligns, fits and plots exactly like a licensed copy. When the thirty days are up it keeps opening and keeps showing you everything — what stops is writing: export and save. Your projects still open.

Download for Windows

Windows 10 1809 or later, 64-bit · 39 MB installer
The beta installer is not code-signed yet, so Windows will show a “protected your PC” notice. Choose More info → Run anyway.

What we would like back

Every calculation in this tool is checked against a closed-form answer — the hashes against their published test vectors, the crossover summation against the algebra, the limiter thresholds derived twice by different routes.

What has not happened yet is the part only you can do: nobody has taken a filter this program designed, loaded it into a processor, played it through a real loudspeaker and measured the result.

If you do one thing with the beta, do that. Design a filter, load it, re-measure, and send back the predicted curve next to the measured one. That single plot is worth more than any amount of arithmetic.

Design the filter.Then print the pagethat goes in the rack.