The Importance of the Small Difference

During a recent visit to Canada's National Research Council, I noticed stuck to the wall of the prototype IEC listening room a page of results from one of Floyd Toole's seminal papers on the blind testing of loudspeakers. The scoring system was the one that Floyd developed, and that we adopted for Stereophile's continuing series of blind tests. "0" represents the worst sound that could possibly exist, "10" the perfection of live sound—a telephone, for example, rates a "2." The speakers in Floyd's test pretty much covered the range of possible performance, yet their normalized scoring spread, from the worst to the best, was just 1.9 points.

Some might conclude from this that the differences described in reviews are in general too small to be worth bothering about. Indeed, my recent cruises in cyberspace reveal that this question is the crux of the debate both between so-called "objectivists" and "subjectivists," and between "audiophiles" and the public:

James Walley on The Audiophile Network, July 8, 1994, after he had carried out A/B/A comparisons on CD players and processors at matched levels, as described by Robert Harley in his "Critical Listening" article in the July Stereophile: "I can't tell any differences!!!!...Well, actually, not really. I did hear some differences. But not the ones I would have expected, nor in every case."

E. Brad Meyer on Compuserve's Consumer Electronics Forum, August 2, 1994, commenting on last month's "As We See It": "I believe...that any difference between components that can't be heard in a controlled certainly not very important."

Jack English on The Audiophile Network, April 21, 1994: "In spite of our use of language, the reality is that many differences which are audible are of nearly infinitesimal magnitude. The more resolving the system, the more likely such an effect is audible. But how big of an audible difference is meaningful? To whom? For how much?"

While most "controlled" tests are too insensitive to detect real differences between audio components, I have no argument with JE's statement that many of the differences we write about are small. When you play this issue's splendid "Recording of the Month," for example, the sound is recognizable as being Mahler's Symphony 1 whether you play the CD on a RadioShack portable or on a $24,000 Mark Levinson transport/processor combination. Within the universe of possible differences in sound, $20,000 buys you a relatively small change.

But does "small" mean "inaudible"? For a long time, I've felt that the difference between an "objectivist" and a "subjectivist" is that the latter has had at one time in his or her life a mentor who could show them what to listen for.

This point becomes paramount when the subject of data reduction is considered. I suspect that the audible degradation introduced by the ATRAC algorithm—and remarked upon in this issue by RH in his review of Sony's second-generation MD machines—is small enough that it would probably go unnoticed by 95% of people. It's ironic, then, that the development of perceptual measurements for the assessment of data-reduced media will shine light on the correlation between what's heard and what's measured. Now that Stereophile has purchased a 24-bit-capable hard-disk editing system from Sonic Solutions to edit our recordings, an important project of mine has been to use that system's capability to subtract one channel from another to investigate the true difference between an amplifier's input and its output using music program, and to look at the hypothetical audibility of the signal residue when a mathematically amplified version of the amp's input is subtracted from its output. This, I suspect, will be a most revealing test.

Does the fact that an audible difference is small mean that it is not important? It seems self-evident that the value placed upon a small but real sonic difference must be an individual one. Take cars: Would you say that a Mercedes represents a sensible if expensive purchase, while to buy a Ferrari is pure self-indulgence, the high price purchasing only illusory improvements in performance? Or would you say that the differences between either luxury automobile and a Hyundai are not important when the cheap car gets you to the mall in about the same time, and uses less gasoline to boot?

Because the answer to those questions will be different for every one of us, for a writer to take upon himself the decision regarding what will be important for his readers is simply arrogant. By contrast, in Stereophile I intend reviewers to describe what differences exist between components and to discuss whether or not such differences represent steps toward or away from higher fidelity. It is then up to each reader to decide for him or herself whether the sonic benefits a) exist, and b) are worth the price, so that they can make their own decision whether or not to buy those components.

The real question is whether it is worth designing and manufacturing components that only a favored few might be able to distinguish. But that's the raison d'être of the High End.