Wine and song
Rick Nelson, Chief Editor -- Test & Measurement World, 11/1/2004
Perhaps nowhere does the test-and-measurement discipline become more contentious than in the acoustic ratings of high-end consumer audio equipment. That's not to say the attendant measurements are more difficult than, for example, characterizing PCI Express lanes. But at least with PCI Express, standards dictate acceptable performance levels.
In the high-end audio field, customers, reviewers, and manufacturers might all consider themselves experts, but none tend to have concrete data. That creates an environment in which savvy marketers induce the enthusiast to spend thousands of dollars on speaker cables in an effort to avoid hearing the annoying clip-clop of electrons hopping from conductor to conductor in cheap stranded wire. That such effects fail to register in laboratory tests is of no consequence to the true believer, who hears them nonetheless.
Now, though, engineers at Harman International are bringing the scientific method to bear on acoustic measurements—at least for loudspeakers. They contend, first, that what you hear you can measure, and that laboratory measurements can predict whether you'll like what you hear. That applies to listeners in general who, in blind tests, far from exhibiting strong individual tastes in loudspeakers, consistently prefer speakers that most faithfully reproduce the original recorded sounds. Unlike preferences for wine, loudspeaker preferences aren't a matter of personal taste. With wine, no benchmark Chardonnay exists in nature for comparison purposes. In contrast, for loudspeakers, there is a benchmark—the original Pavarotti aria, for example.
That fickle humans provide consistent and repeatable results in psychoacoustic experiments isn't too surprising. The general concept has been well known since publication in 1933 of the Fletcher-Munson curves relating loudness (in phons) to frequency. Perhaps the most contentious finding of the Harman team is that trained listeners' results correspond with those of untrained listeners. In one precedent—the development of the scale defining pitch in mels—trained listeners had to be avoided. Asked to adjust an experimental tone to twice the pitch of a control tone, a trained musician would automatically choose the octave.
Nevertheless, the Harmon team has the data that demonstrates trained listeners' efficacy. For details on the Harman team's efforts, see p. 24. And if you're the enthusiast mentioned in the second paragraph, don't forget to send me your defense of $100/ft cables and your detailed regimen for breaking them in. Go to www.tmworld.com/audio to respond.
Contact Rick Nelson at rnelson@tmworld.com.
