November 1, 2003The Mêlée Over Measurements
It may come as a relief for some people to know that most audiophiles, in my humble opinion, are not nerds. They are actually hedonists, gluttonously devouring the pleasures of the ear while constantly pondering their next step on the path to upgrade enlightenment. Unlike the small minority of genuine audio nerds, for whom the "wowee" factor of the technology is critically important, most audiophiles consider their gear to be a means to achieving better sound.
Still, we want to be scientifically minded. Most of us like to think were well-versed in the technology, and that our components are rationally designed and selected in a way that will enhance the audio experience. In keeping with this, many people select gear, at least in part, on the basis of specifications and measurements. The well-known fact that standard measurements correlate poorly with sound quality flies in the face of our collective self-image, and is therefore usually ignored.
Nevertheless, the possibility that the differences that people hear between components may not only be immeasurable, but might not even exist, hangs in the air like a bad smell at a cocktail party. Everyone pretends its not there, but there it is. To make matters worse, every so often some conflicted souls id gets the better of his ego, and he publishes a tell-all "scientific" report that claims to show, by means of real, live measurements -- expressed using actual graphs and charts, no less -- that the differences we hear do not or cannot exist.
Take Phil Wards "Running on Empty," in the recent issue of Hi-Fi+ (No.23, pp.33-37, May 2003). Using two pairs of B&W CDM1 NT bookshelf speakers, Ward shows that neither the frequency response nor the damping factor (the spectral decay) changed when one pair of speakers was thoroughly run-in and compared with a pair that had not been run-in. He concludes that the subjective effect of break-in must be "psychoacoustic." That is, you get used to the sound to the point where it seems that the run-in speaker sounds better. To put it more bluntly, hes saying that its all in our heads.
With the greatest possible respect to Phil, my reaction can be summed up in three words: horsefeathers, poppycock, and twaddle. There are many scientific papers in peer-reviewed journals in which the authors measurements fail to correlate with or capture a well-known phenomenon. The next time you see a report like that, in which the author goes on to conclude that the phenomenon must not exist, let me know so I can cancel my subscription.
Not that the measurements themselves are necessarily invalid. Im prepared to accept them, and I certainly find it interesting that they remained relatively constant throughout the run-in period. But what if frequency response and damping factors are simply not responsible for the sonic effects of break-in? In this case, as in general, absence of evidence is not evidence of absence.
Im reminded of a story from Doug Schneider, the publisher of the SoundStage! Network. As measured in the anechoic chamber at the National Research Council in Ottawa, Canada, one speaker had a 3dB hump in frequency response across the midrange. Despite the fact that 3dB represents a doubling of subjective volume, the hump was detectable on close listening as no more than a slight degree of extra midrange presence. My conclusion? Frequency response may be a very weak predictor of subjective sound quality. As anyone knows who has tweaked the knobs of an equalizer, you can get away with a variety of deliberate frequency-response perturbations that still sound like music.
Finally, with regard to Phil Wards article: In the end, this avenger of objectivity did not even bother to report whether the run-in speakers actually did sound better than the non-run-in pair. If they did, then the psychoacoustic theory might need some fundamental work.
Ultra Audio is part of the SoundStage! Network.