Measurements & Sound Quality

"One of the worst-kept secrets in audio engineering is that what we hear does not always correlate with what we measure." So wrote the late Richard Heyser 30 years ago, as quoted in Time Delay Spectrometry, a 1987 anthology of his writings (footnote 1). What do we hear? Music heard live consists of a sound pressure that changes according to the logical demands of two things that have no physical reality: the way in which music is structured in time and pitch, and how that structure is ordered by the composer/musician. Heyser, one of the most perceptive audio engineers I've had the privilege to meet, repeatedly emphasized in his essays and papers that the reproduction of music is a multidimensional event.

According to Heyser, the real aspects of the concrete framework that supports the two abstractions are at least five-dimensional. Sound has a "where," which covers three dimensions by itself. It has a "tone," which includes pitch and timbre, themselves independent variables. Its intensity, a "how much?" that varies with time, represents dynamics. It has a "when" aspect in that the listener's instantaneous perception of musical values depends very much on what has gone before. And which of these aspects is the most important when assessing quality will be different for each listener.

By contrast, any typical measurement is a two-dimensional, or at best three-dimensional reduction—as in Stereophile's loudspeaker waterfall plots, where amplitude is plotted against time and frequency or direction and frequency. And the choice of which to plot against what may not be related at all to sound quality, but to the practicalities of being able to perform the specific test. There's the old story of the drunk looking for his keys under a street lamp. A passerby joins in the search and, after a fruitless few minutes, asks where the drunk has dropped them. "Over in the bushes," answers the drunk, "but it's too dark to look there."

To make things worse, every perceived aspect of a component's sound quality is affected by more than one technical aspect of its design. If a listener describes a loudspeaker as sounding "bright," for example, is it because the speaker has a tilted-up on-axis frequency response? Or is it because the speaker may be flat on-axis but its dispersion has some mid-treble peaks? Or because the power response tilts up? Is it because the speaker has excessive midband distortion? Is it a combination of all four, exacerbated by a lean-sounding woofer tuning? Or does the brightness have nothing to do with the loudspeaker's "sound," but is instead due to the partnering amplifier being driven into momentary, music-dependent overload by a speaker impedance that features a punishing combination of low magnitude and extreme phase angle? Although any single measurement will always have some connection with sound quality, that connection can be tenuous.

Despite Mr. Heyser's wise words of warning, I announced in August 1989 that Stereophile was introducing a program of measurements in support of its subjective equipment reports. "[We] will routinely measure every component submitted to the magazine for review," I wrote, "taking into account the most up-to-date research...." Our goals were threefold: to show up faulty components before we sent them out to the selected reviewer; to ensure that there wasn't some simple reason for a component to sound the way it did; and to build up a database that would eventually throw up correlations between what is heard and what is measured. The measurements in Stereophile reviews would be used to try to explain the "why" behind "what" was heard.

As the magazine continued to add standardized measurements to its reviews, I had to pen various cautionary notes in response to criticisms from both "objectivists" and "subjectivists": "Those who place their belief in measurements alone should remember that it is still the experience itself that matters, not the description of the experience, no matter how thorough or well-researched that description," I wrote in November 1990, adding in February 1996 that "The integration of measurements into Stereophile's equipment reviews is not to describe or replace the listening experience—that is, and probably will always be, impossible. Without listening, there is no way, for example, of measuring something as universally perceptible as the quality of a stereo soundstage."

While we have made considerable progress with the first two of our 1989 objectives, it is fair to say that Stereophile is still a long way off from achieving the third and most important goal: using the database to examine what kind of measured performance good-sounding products have in common. Some patterns indicating a causal connection between what is heard and what is measured have emerged from the mountain of data, particularly regarding loudspeakers. But with amplifiers and digital products, the more we find out, the less we seem to know.

Which is why I am glad to welcome the English audio engineer Paul Miller to our staff as "Test & Measurement Consultant." (Tom Norton's review of the Kinergetics KBA-280 amplifier in this issue sees the first measurement taken with Paul's "Amplifier Profiler.") Paul tells me that his goal is "to develop test and measuring equipment to take us into the next millennium, to try to determine what is actually going on." Stereophile intends to be at his side when he does so, remembering that while good measured performance is important, this will be always subsidiary to good sound.

Footnote 1: Published by and available from the Audio Engineering Society, Inc., 60 East 42nd Street, Room 2520, New York, NY 10165-0075; $27 for AES members, $30 for nonmembers. The AES website offers a secure transaction page for credit-card orders.