To achieve the fundamental goals of sound reinforcement, speaker designers invariably turn to HF horns for their significant practical advantages. Horns project energy into a defined coverage area, and so help to control reverberation and interference. They are much more efficient than direct radiators, making it easier to achieve high peak SPLs. Properly designed, horns also add well in arrays, providing smooth, predictable coverage.

But horns also have limitations, notably in their sonic characteristics. Every audio professional is all too familiar with the "horn sound"—a megaphone-like "honkiness" with smearing of transients. When horns combine with cone LF drivers in a full-range system, the disparity between low and high frequencies can destroy natural timbres, creating a false, "amplified" quality with an obvious break at crossover.

Can a PA loudspeaker ever sound as good as a studio monitor? This is the challenge that EAW's Director of R&D, David Gunness, posed when developing the NT Series. The first step was to make high-resolution measurements of a proposed two-way system, with the goal of identifying problems that would yield to correction —that is, that did not vary with signal level or measurement axis. The process revealed distinct, correctable anomalies that were traceable to basic loudspeaker physics.
This spectrogram shows the on-axis measurement data filtered to isolate the 1 kHz to 10 kHz region. Notable in the display is a series of energy peaks at intervals of less than 0.5 ms. This prominent resonance, caused by reflections between the compression driver diaphragm and the horn mouth, is responsible for the characteristic horn “honk.”
This spectrogram shows the data filtered to isolate the octave from 10 kHz to 20 kHz. Here, we see multiple arrivals along different propagation paths through the phase plug, appearing as a series of very high-frequency peaks occurring primarily within the first millisecond. This energy causes transient “smearing” that affects the quality of impulsive sounds and degrades imaging. Universally ignored in PA loudspeaker designs, problems in this octave contribute substantially to the artificial sound quality.
Turning to the critical midrange, filtering to isolate the response below 10 kHz reveals delayed energy in the region between 300 Hz and 3 kHz that extends as late as 8 ms. These resonances, caused by reflections across the LF driver cone surface, introduce midrange coloration — especially in systems with a 15-inch LF driver.
Dave Gunness knew he needed very precise signal preconditioning to correct these problems. Turning to digital signal processing, he worked with EAW partner Acuma Labs to develop a proprietary matched Z transform that produces a stable filter while properly preserving the resonant frequencies of decay trails - and is thus optimally suited for speaker correction. This spectrogram shows the result as implemented in the NT Series: a near-perfect impulse response. Named Gunness Focusing in recognition of its creator, this sophisticated DSP correction gives the NT Series its phenomenal accuracy and clarity.