This series of loudspeaker articles I’m writing each month wouldn’t be complete without a discussion of measurements -- frequency response, impedance, distortion, etc. -- which is why I’m bringing up the topic early. Measurements are important for two reasons: first, they’re vital to understanding high-performance loudspeakers; second, measurements are a contentious issue among audiophiles and manufacturers, one that spurs endless debate about their worth.
I’ll never forget the day, about ten years ago, when we announced that we were to begin testing loudspeakers at Canada’s National Research Council (NRC). Within minutes of the announcement, I received a call from the president of a speaker manufacturer. First he congratulated me on taking such a bold step in the online publishing business. He was happy that someone was going to publish something objective and credible about high-performance speakers that would help readers separate the wheat from the chaff. Then he told me, flat out: "But you’ll be out of business in six months."
He felt that while published measurements would be a great thing for the world to see and would bring us tremendous credibility, most manufacturers would be too scared to have their speakers subjected to rigorous measurement, which would make many of them look like crap. As a result, they’d refuse to send us speakers or otherwise support us, and our publication would die. That hadn’t occurred to me when we signed the measurement contract with the NRC, but I so firmly believed that what we were doing was right that I decided to wait and see what the results were.
That speaker maker’s prediction was partially correct. A few companies did refuse to send us speakers to review and measure, and most of those companies also stopped supporting us. More than likely, their speakers would have measured like crap, and that’s why they took that stand -- or, rather, ran scared under the pretext of taking a stand. Even today, a few manufacturers balk at the idea of their speakers being measured.
But that speaker manufacturer was wrong about us going out of business for lack of support. In fact, the opposite happened. We were swamped with requests from credible speaker companies who wanted their speakers reviewed and measured in such a world-class facility as the NRC’s. Those manufacturers who were eager for this to happen far outnumbered those who didn’t. As a result, measuring loudspeakers and publishing the results not only didn’t put us out of business, it increased our business. That’s one I never saw coming. But even now, after we’ve published more than ten years’ worth of measurements done at the NRC, the process is not without controversy.
Many of our readers know how to read and interpret the measurements and put them to use in shopping for speakers -- they often thank us for providing them with information that they consider of value. In many instances, good measurements have shown a strong correlation with good sound. And almost every one of the very best-sounding speakers we’ve reviewed has also measured well. (I’ll discuss what makes for "good" measurements in future months.)
On the other hand, many people have questioned the validity of measurements, particularly in the case of a speaker that they think sounds good but that measures quite poorly. They don’t think the two things have anything to do with each other, even if the discrepancy can be easily explained. Instead, they throw the baby out with the bathwater by dismissing measurements altogether, and often, on the various audio forums, lash out about how speaker measurements are useless. Basically, they take this stand out of ignorance, which is why the topic of measurements seems to be a constant source of unresolved debate.
I could go on about why some audiophiles get so stirred up over measurements, but Jeff Fritz has already done a really nice job of explaining it, in his September 2010 editorial "Why Some Audiophiles Fear Measurements," which appeared on Ultra Audio, our sister site devoted to two-channel audio. Read it -- he says a lot of stuff I would here, if it weren’t for the fact that that would take too much time and focus away from what I want to talk about in the rest of this installment: how measurements are used, and by whom.
Obviously, I believe that measurements matter. But it was fellow SoundStage! Network writer S. Andrea Sundaram who first pointed out to me that measurements are most relevant to designers, mostly because measurements help them work more efficiently. While it’s possible to design a loudspeaker without taking measurements, it’s much easier to do so with them. That’s because we now have decades’ worth of research that correlate certain objective measurements with subjective listening impressions -- there’s not as much black magic or mystery to designing a good loudspeaker as some people think. Leveraging this measurement-based research in conjunction with carefully implemented listening sessions results in a far more efficient use of designers’ time. In contrast, trying to accomplish the same thing without taking any measurements is a bit like hunting for something in the dark -- you might eventually find what you’re looking for, but it will probably take a very long time.
Measurements are less important to reviewers, to whom the final sound of a loudspeaker is what most matters. But measurements can still be quite valuable to reviewers if they’re skilled at reading charts and interpreting the data presented there. I use measurements to tell me what the speaker will generally sound like -- frequency-response graphs taken on and off axis are crucial for this -- but I also use them to assess if the speaker has been competently designed. Because some readers might not realize that reviewers use measurements in this way, this is the area of design competence that I’ll focus on for the rest of this piece.
When I look at the measurements we take at the NRC, I ask myself: Is there method to the madness? Or, maybe it would be more accurate to say: Is there method at all, or is it complete madness? Basically, the various measurements can disclose quite handily whether the speaker was engineered by a designer who knew what he was doing, or was the result of guesswork by someone basically jamming drivers and crossover components into a box. I look at the speaker’s frequency-response chart and assess the overall shape or trend, as some designers like to call it, of the graph. This trend can tell you quite a bit about the sound, including if the speaker has been deliberately voiced by the designer -- in other words, engineered to sound a certain way. For instance, when I looked at the frequency response of the Revel Ultima Salon2 (a speaker that I reviewed in December of 2009), it was obvious, from the smoothness of the FR lines both on and off axis, and the way that they traveled from the bass through the mids and the highs, that Revel’s designer(s) were aiming for a specific target response. Nor is there anything in any of the Salon2’s measurements that indicates that the designer(s) did anything seriously "wrong." In other words, the Salon2’s measurements are those of a competent design realized through credible engineering.
Ditto for the Vivid B1, Paradigm Reference Signature S8, and PSB Synchrony One -- one look at their FR curves and it’s obvious that these speakers have been thoroughly engineered, not merely thrown together. That’s not to say that these speakers all sound the same -- far from it -- only that there’s method, not madness, behind their creation. For me, that’s important: Before I endorse a product, I want to know that a certain level of rigor and competence has gone into its design. Measurements help tell me that.
Ironically, measurements are least important for consumers, who mostly just want to buy an audio product that sounds good to them and is reliable. I call this ironic because we publish our measurements for consumers to read. But that doesn’t mean that measurements are entirely irrelevant to consumers. There are a few reasons measurements are still important to them, even if the reasons to use them are far fewer than for reviewers and, especially, designers.
First, as I mentioned above, technically savvy consumers who know how to read and interpret measurements can use that information to make better buying decisions -- like the technically competent reviewer, they’ll be able to glean from measurements something about a product’s sound, and discern if it’s been competently designed. Second, measurements help keep manufacturers honest by publicly disclosing objective data determined by a third party about how their speakers perform -- data that can be used to verify the manufacturer’s claims of the speaker’s sonic and technical performance. If no one published such measurements, manufacturers could make any claims about their products they wanted to in order to make them appear to be better than they are, and readers would never know better. Third, even if consumers have no idea how to read and interpret measurements, just having them there gives them the opportunity to learn how to use them. I know that when we began publishing measurements, few readers -- and few reviewers -- understood the importance of many of them, and couldn’t relate them at all to a product’s performance. Ten years later, many more people know how to interpret and use that information to make more informed buying decisions. I hope that in another ten years there will be many more.
Measurements scare the hell out of some people -- mostly, manufacturers who make shoddy products and consumers who want to pretend that correlations between objective measurements and listener preferences don’t exist. But in my opinion, measurements are extremely valuable for designers, quite important for reviewers, and fairly relevant to consumers.
If you’re one of those people who fears the measurement process, please don’t. Instead, embrace them, learn about them, and you’ll be able to make more informed buying decisions. Remember, measurements didn’t put us out of business; they grew our business. There must be something to them after all.
. . . Doug Schneider