Why we would never use a DEQX or similar device to build a loudspeaker !

[Tranquility Bass]

Don't be fooled by the blurb on the DEQX website regarding group-delay correction or compensation of loudspeakers. It's a way oversimplification of reality and they have been pushing the same barrow for decades which is probably why they have never commercialized their own speaker systems using it, instead pushing the gimmicks onto other naive loudspeaker designers. Trying to fix up a loudspeaker on a single axis was never going to work. Loudspeakers are far more complex than this and radiate into free space in all directions and not just on one axis alone ! What would fix up a signal chain on an electrical connection simply won't work with a distributed device such as a loudspeaker. This is why no serious hi-end audio manufacturer would ever use a DEQX as well ! Yes, sure our own  UPP can do long-length high sample rate, ultra deep convolution and FIR filtering and do a much higher resolution correction than a DEQX could ever do but do you really want to go down this path ?  When it comes to loudspeakers there simply is no free lunch here and no easy fix and throwing lots of DSP at the problem can often make the problems a lot worse. It certainly is not a panacea for a bad speaker design to begin with !

Anyway don't take our word for it but we suggest you read these three important documents before outlaying your hard earned money on such a device or anything that is just a fancy inverse convolution filter ! These are documents written by other well established audio engineers and echo the same sentiment as us. This is what DEQX will never tell you and why they never sell loudspeaker systems themselves using their own DSP hardware and even if they did it was always going to be a band-aid solution to a complex problem !

https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf

And the take-home message from this document is:-

DSP loudspeaker crossovers done right

From the above we've learned that:-

• Heavy-handed correction exacerbates acoustical problems
• Sharp, linear-phase filters cause pre-ringing
• Targeting an exact linear-phase sum can cause pre-echos.

In short, brute-force correction sounds grainy and smudgy. When you hear cymbals go "splash" instead of
"crash", it's naive DSP at work. So:-

• Do not shave off the hair, a nasty stubble will grow back.
• Do not correct beyond the very beginning of the impulse response.
• The gentler you correct, the wider the angle over which the correction still improves things.
• Target a minimum phase sum.

For the time being I would strongly recommend designing the correction manually. This rules out FIR as the
main workhorse. For each bump or dip one corrects, one should know exactly where it comes from, and make
sure that it isn't better corrected for acoustically. Unfortunately, designing DSP filters does not relieve one from having to know one's acoustics.

Also:-

https://linea-research.co.uk/wp-content/uploads/LR%20Download%20Assets/Tech%20Docs/CrossoverFilters%20White%20Paper%20-C.pdf

Also shown in Figure 8 is the impulse responses for a complementary high-pass brick-wall FIR. If we construct a crossover filter bank with such a complementary pair of filters, the Gibbs ripples are also complementary (since we will expect the filters to sum to a flat response with linear phase, producing a perfect impulse). The summed result will thus be free of any Gibbs ripple, so what's the problem? The problem is off-axis. The complementary ripples will only cancel if the delay suffered by the signal from each driver is identical. Off-axis, where the path lengths differ, the ripples will not cancel, leading to the possibility that Gibbs ripple might become audible (just like a high-Q ringing filter).

Such summing errors will be more pronounced at higher crossover frequencies because the ripples are more closely spaced. Lower frequency
crossovers will have wider ripples which will more easily cancel in the presence of off-axis induced delays.

It is evident that steeper cut-off slopes give rise to Gibbs ripples of greater duration. It makes sense therefore to restrict the cut-off slope to be no more than is necessary for the application.

And from the Linkwitz Lab website pretty much the same cautionary tale

https://www.linkwitzlab.com/frontiers.htm

I - Digital crossovers

Some people think that digital crossovers will replace analog ones, because digital filters can be designed with desirable characteristics that are impossible to realize with analog circuitry. In particular, lowpass and highpass filters with extremely steep slopes and linear phase shift are possible. Steep slopes reduce the overlap region between drivers. Linear phase shift eliminates waveform distortion and merely causes a delay of the signal.  Such characteristics can be obtained from the digital equivalent of tapped delay line filters, which have a finite impulse response (FIR) duration that depends upon the number of taps used. Digital FIR filters can have almost any desired frequency response, if the number of weighted taps is made sufficiently high. [1]

The linear phase shift comes at a price. The impulse response rings. The more so, the steeper the filter slopes. Both lowpass and highpass sections of the crossover ring, but when the outputs are combined, as for a crossover, then the two impulse responses add to a non-ringing, delayed pulse.

All would be fine, if we listened only in anechoic spaces or to speakers with coincident drivers. In reality we use speakers in rooms with reflections and reverberation and the the drivers are separated from each other due to their sizes. As a consequence the off-axis response of the speaker matters and contributes to what we hear. With the drivers non-coincident, the lowpass and highpass outputs are delayed different amounts at points off-axis, and the ringing is no longer canceled in the addition. In the best case the drivers might be coaxial, but this has another set of problems. Very steep crossovers can also cause a very abrupt change in the polar pattern of the speaker, when transitioning from a large diameter driver to a small one. Under reverberant conditions and/or listening off-axis this may have audible consequences.

[Tranquility Bass]

And from DEQX's own white paper !!

In consumer speakers where crossover slopes of typically 12dB/octave are used, the same sound is coming from both drivers, typically for an octave above and below the crossover frequency. Unless these sources are perfectly timed to meet in phase with each other (time aligned), the resulting mix of sound will produce 'comb filtering', which can sound 'nasal' or 'flanged'.

Another problem has also been introduced by using multiple drivers to reproduce sound – a single instrument may now sound like it is coming from different places. This results in poor off-axis performance because varying degrees of summation and cancellation occur depending on the relative distance between each driver - a phenomenon known as 'lobing'. Because the woofer is often physically placed further back than the tweeter, this lobing is also non-symmetrical in the vertical plane. The loudspeaker's off-axis performance is important because although this is not necessarily what is heard directly, it is ultimately heard because of its coupling into the room.

THE DEQX SOLUTION
The active DEQX crossover and correction processors provide help to resolve the problems mentioned above at a number of levels. Although impractical even in 'active' analogue crossovers, DEQX provides true linear phase very high-order crossovers that provide absolutely precise time alignment and phase coherence between drivers, while limiting the 'bleed' of similar frequencies over the crossover region so that lobbing effects occur only at the imperceptibly narrow crossover region, and dispersion patterns remain similar between drivers.

Yes Mr DEQX, the off-axis response is very important especially when trying to deploy steep slope linear-phase crossovers. Well and truly busted Mr DEQX

[Tranquility Bass]

From their latest facebook post just more of the same. Old habits die hard