Priority Shift
What makes a loudspeaker “good?” What’s an example of a “good sounding” loudspeaker? Of course, if you ask ten different sound engineers, you’ll get eleven different answers. At the heart of this question is a mixture of different qualities, both subjective (“I find the tonality pleasing”) and objective (“This loudspeaker has very low distortion.”) As discussions of this nature arise frequently amongst my colleagues (consider it an occupational hazard), I began to notice some interesting trends.
When I ask a systems engineer or system designer why they prefer certain loudspeaker products, they usually mention technical or logistical aspects - the weight, the rigging hardware, the functionality of the prediction or control software - in other words, things that have nothing to do with the sound of the loudspeaker, but everything to do with their day-to-day jobs and interaction with the product.
When I asked FOH engineers the same question, many of them told me they liked one brand or another because they “don’t have to do much to it.” After some elaboration, what many of them would tell me is that they like the “out of the box” tonality achieved by a certain system, in that they don’t have to apply much EQ to the system to get it it to sound the way they want.
Loudspeaker manufacturers are undoubtedly asking end users the same questions, because the “listen to how good it sounds without EQ!” point is one I’ve encountered time and time again, in loudspeaker demos, tradeshow booths, and marketing materials. Clearly this is a selling point.
Now let’s be clear - certainly the manufacturer’s tuning that’s “baked into” the loudspeaker preset should result in a loudspeaker that accomplishes what I’ll call a “useful” tonal response without tinkering from the end user. Not missing a whole bunch of 250 Hz, no annoying narrow-Q peak at 2.5 kHz - that stuff should be dealt with in the manufacturer’s preset. And by and large, it is. DSP is a vital ingredient in the modern era of loudspeaker design, which means manufacturers can design a loudspeaker to achieve the acoustic and mechanical performance they desire, and then use the processing preset to shape the resulting response into their desired tonality.
The red trace at the top of Figure 1 below shows the raw acoustic response of a single element of a popular large-format line array product measured from about 10 feet, without the manufacturer’s preset in place in the amplifier. The green trace shows the response of the filtering applied by the manufacturer’s preset for that loudspeaker, and the bottom pane of Figure 1 shows the resulting response in blue. (Some ripple is visible in the low-mid range due to the concrete surfaces of the warehouse in which the measurement was taken, but the point stands.)
FIGURE 1 - Top: Acoustic response of a single element of a popular large-format line array product (red), and the electronic response of the manufacturer's loudspeaker preset in green. Bottom: the combined response as experienced by a listener.
The Quest for Tonality
But besides accounting for all the high-Q peaks and dips in the acoustic response and creating a generally useful resulting response, how much effort should manufacturers be devoting to this “perfect out of the box” tonality?
I might argue that such an achievement isn’t necessarily even realistically achievable - for a number of reasons. First of all - “perfect” tonality is subjective. Perfect to whom? Sure, there’s a “center of the bell curve” in terms of an overall tonal contour that seems to work well for mixing live music, but many of the FOH engineers I work with regularly have specific preferences for target curve - and they’re all different. So making a loudspeaker that hits a certain person’s target curve perfectly by default means others will still need EQ to adjust it to their liking.
Then we have the program material. A target curve that works well for corporate, speech and musical theater might be disastrous for an EDM festival. And of course we haven’t even begun to consider the list of other variables that affect the perceived tonality of a sound system - line length, box count, proximity to other loudspeakers, walls, and environmental boundaries, the reverberant characteristics of the space…
This, to me, makes the whole “let’s listen with no EQ” demo quite a bit superfluous, even though multiple major manufacturers have tried to impress upon me this very point. This once crossed into absurdity when a manufacturer tried to convince me that it was “very important” that I apply no EQ to their system when aligning it for an event, to impress the show promoters and talent buyers in attendance. I tend to think a promoter would more likely be impressed by a PA that sounds excellent, rather than a PA that sounds “pretty good” plus the knowledge that no EQ was used to get it to this “pretty good” state. (None of the promoters in attendance asked to look at my processing settings.)
The reason I find all this to-do about stock tonality so non-stimulating is the simple fact that it’s really easy to change if we don’t like it. We have EQ for that. In addition to your garden-variety parametric and shelving filters, most manufacturers also include proprietary filters to change broad-strokes system tonality with a few clicks (U-Shaping, Array Morphing, CPL, etc). No one is hard-up for filters these days. It’s an easy fix.
Warts and All
The elephant in the room is that many loudspeaker products do ship from the factory with other attributes that may not be so desirable, but unlike tonality, can’t be changed by the end user, and can have a considerable effect on the sound of the system. A big category is linearity and distortion. I am very glad that in the modern era of loudspeaker design, we seem to be trending away from loudspeakers having “a sound” and towards increased linearity and high fidelity transmission. Current-generation loudspeaker products, almost without exception, have significantly lower THD (Total Harmonic Distortion) than their comparable previous-generation products. This is a good thing, and I can say that I have certainly become more sensitive to distortion products in older loudspeakers the more I listen critically to newer, more linear products.
(Some mix engineers - and indeed many listeners - enjoy the sound of certain non-linearities in musical content, such as the pleasing distortion characteristics that are hallmarks of analog tape, vinyl, tube saturators, and so forth - but those are artistic choices that belong in the mixing console, not forced upon us by a sound system struggling to maintain linearity.) The loudspeakers are part of the transmission chain, and in a typical system, contribute more distortion and non-linearity to the signal than all the other system components combined. If we don’t like the distortion levels of a loudspeaker, there’s nothing we can do as an end user to fix it, other than use a different loudspeaker.
Closely related to the subject of linearity is the phase response of the element. Put simply, the wider the frequency range having a flat phase response, the more of the spectrum that is leaving the loudspeaker at the same time. Remember that the magnitude and phase responses are the frequency-domain equivalent of the Impulse Response, and a flatter phase response translates to less “smearing” in the IR.
Figure 2 below shows the out-of-the-box responses for three large-format line array systems from common manufacturers, each 8 box-hangs with the same splay angles and same trim heights, measured from about 50 feet. Without applying any EQ, and disregarding the difference in LF extension of the purple system, we can see that they achieve extremely similar tonal responses (bottom pane).
FIGURE 2 - Measured responses of 3 8-box large-format arrays from common manufacturers with same splay angles, before EQ.
The phase responses, however, are very different. The blue system wraps around 80 Hz then stabilizes and remains phase linear all the way to the top of the spectrum. The red system wraps at 250 and then gently ramps from 180 to 0 throughout the remainder of the frequency spectrum. The purple system wraps multiple times all the way up to 2 kHz.
Despite having very similar tonalities, these three systems will sound very different for a variety of reasons, including the THD as discussed above and the polar pattern as discussed below, but in general, newer loudspeaker products are trending towards more linear phase responses and the more controlled IRs that come as a result, which I find to be a welcome trend.
Spin Me Right Round
Similarly, the polar behavior of a loudspeaker cabinet can have a profound effect on not only the listening experience of the audience, but the acoustic environment on stage. We can stand in front of a loudspeaker and adjust the EQ until we like it, but we can’t control how consistent that tonality is for listeners located elsewhere in the coverage pattern - near the edges of the horn, for instance - or indeed what’s coming off the back of the PA and onto the stage.
We can use AFMG’s free EASE GLLViewer program to import the .GLL acoustic prediction libraries provided by most loudspeaker manufacturers and render some graphs that will allow us to examine the polar behaviors of different loudspeakers. GLLViewer can generate an impressive variety of different graphs from the raw .GLL data but for our purposes here, we will investigate how the pattern control behaves in the horizontal plane.
Imagine drawing a 20-foot wide circle in chalk on the ground and marking off the perimeter of the circle, like a clock face, in 10-degree increments. Place a loudspeaker at the center of the circle and measure its response from each location around the circumference. That’s effectively what the charts below show - a transfer function overlay showing the magnitude response at a distance of 3 meters, measured every ten degrees moving horizontally around the circle to a maximum of 100° off-axis. (GLL data is often generated by placing the loudspeaker on a turntable and rotating it while the mic remains stationary, sometimes referred to as a “spin-o-rama.”)
The chart is normalized on-axis, which means it uses the 0 degree (center position, on-axis) measurement as the reference, and normalizes all the traces to that, so the other traces show us how the response changes as we move horizontally through the coverage pattern of the loudspeaker relative to what’s happening on axis.
Figure 3 below shows us a very well-behaved example of a dual-8” line array element. The flat red line is our on-axis 0° reference and as we sweep through the horn pattern and end up 100 degrees off axis, the HF drops more or less uniformly. We can see that the HF hits -6 dB by about 50 degrees, so we would classify this as a 100 degree horizontal dispersion.
FIGURE 3 - Horizontal normalized "spin-o-rama" of a well-behaved dual-8" line array element.
This loudspeaker exhibits an almost textbook high-frequency response, and actually behaves far better than most of its competitors. Let’s look at another example, another dual-8” element, from a major loudspeaker manufacturer. (FIGURE 4).
FIGURE 4 - Horizontal normalized "spin-o-rama" of a less well-behaved dual-8" line array element
This loudspeaker has a very different tonality for the listeners near the edges of the pattern - this element is marketed as 110° dispersion, so the listening positions at the black line (+/- 50°) are fair game. The midrange plunge means that listeners in this part of the coverage hear 10 dB more 1.5 kHz than 1 kHz, and the HF falls off above that, creating a very high-mid heavy response. This translates to tonal inconsistency across its coverage area, and it can’t be fixed with EQ. This is a design issue, not an end user adjustable parameter. Side-pattern radiation also needs to be considered whenever the array is hung next to a reflective wall, such as in a wood-surfaced orchestral hall. The tonally skewed energy reflects off the side walls and back into the listening area. Good polar behavior becomes extremely important whenever sound systems are deployed in reverberant spaces because they will affect the listening experience of all the listeners in the space, and it’s something we can’t fix if it’s a problem. Don’t like it? Need different loudspeakers.
The Focus
There’s no call to action here, and that’s the point - we’ve looked at several important ways in which loudspeaker products can be differentiated from each other, and how those differences might affect our experience as end users and listeners. None of these properties - distortion, phase response, polar pattern - are things that we have control over - they lie solely in the domain of the manufacturers’ R&D teams. So why is so much of the conversation about tonality - the one thing that’s trivial to change?
When auditioning sound systems, most of us tend to pay attention to the on-axis tonality by default. That’s the perceptual “low hanging fruit,” if you will. It’s also, I would argue, one of the least important parameters by which to adjudicate a loudspeaker, simply because we can easily adjust it ourselves. Linearity and polar behavior are important considerations at the product selection phase, since we’re stuck with them, so it may be a wiser strategy to select a product having the qualities that are desired for the use case, with the knowledge that we can apply EQ to achieve whatever tonality we desire.