A seventeen-chapter working journal on 2D and 3D acoustic steering, FIR algorithms, omnidirectional measurement methods, and the slow science of building a loudspeaker that moves without moving.
Outdoors, with no ceiling and no reflective surface for hundreds of feet, the 2D-steered line of drivers points straight up. The microphone sits by its side, in the horizontal plane that cuts across the array.
Each driver is heard exactly ninety degrees off its own axis — in the equatorial plane where its radiation is symmetric of revolution. The full directivity of the line array goes straight into the sky and never returns. That is how, from our own drivers, we build a perfect omnidirectional source — and only then can the Digital Horn be measured with truth.
Putting it all together — a path into an innovative system capable of multi-beams, active suppression algorithms, and massive FIR use. The first steps; the skeleton of the project; the model proposals DHR01 and DHR02.
A letter about beam steering, phased arrays, Huygens' principle and lobes. Questions about the new summator, the paraline, and why only the SBH10 and SBH20 were developed. A secret tool inside Direct is on its way.
An advanced speaker that changes the audience area without any movement — only DSP. The immense, tremendous low profile. Thirty decibels between audiences. Why we must go back to 2D.
Having enough theory — we prepare the first measurements. Outdoors, flat ground, no reflective surfaces in six hundred feet. The array horizontal on a turntable. The microphone by its side. The omnidirectional measurement method, at last.
The inner anatomy of the FIR engine — how a small DSP program, loaded with the right coefficients, becomes a lens that bends sound without glass.
The catalogue: focal, plane-wave, beam-steered, multi-beam, null-forming. Each one a recipe that tells every driver exactly when to speak and how loud.
Several beams at once, several programs at once, from a single body of drivers. The intuition, the mathematics, the first simulations.
Where the algorithm stops being written by hand and starts being discovered — the moment machine learning enters the Digital Horn.
A short, dense page from the lab — measurements against simulation, the first correlations, the first quiet victories.
Twelve megabytes, thirteen figures — the densest single chapter of the diaries. Where the drawings speak more than the sentences.
A long day of work, split in two parts. Part I covers the geometry, the measurements, and the first drafts of new coverage algorithms.
The continuation. Late afternoon into evening — more measurements, more comparisons against simulation, the chapter that would not fit in one file.
A dedicated report to Tom Danley on the end-fired research line: the geometry, the missing information, the drawings needed to fully grasp it.
The trade-off at the heart of every array: how close can the sources be, for how high a frequency, without lambda breaking the picture apart. A study in numbers and pictures.
New measurements, new scripts, the first signs that the FIR generator is becoming a tool that does not need its author to produce a good beam.
Winter comes to the south. The lab is quieter, the numbers tighter — a consolidation entry before the year's end.
Two signals, two directions, one column of drivers. The first open response to the idea of 3D screens of speakers. Eight FIRs, two processors, one passive sum — and an audience hearing a Latin band on one side while Phil Collins plays on the other.