Deeper Meanings Letters part 5

What is good!
Editor: In Vol.13 No.7, Robert Harley wrote a scathing rejection of those who do not share his particular view of audio evaluation. In his closing paragraph he wrote, "now to get back to Dr. [Stanley] Lipshitz's question that prompted this essay: 'Ah, but how do you know what is good?' The question was clearly not meant to be answered."

I have heard the description "warm" approvingly used when the sound was the result of the obvious nonlinear distortions produced by vacuum tubes. Furthermore, I am totally repulsed by what members of the German SS regarded as "good."—Loy Weiss, Berkeley, CA

Stanley Lipshitz responds
Editor: With reference to Robert Harley's editorial (pp.5-15) and comments (pp.43-51) concerning what I said at the Audio Engineering Society's recent 8th International Conference: "The Sound of Audio," as reported in Stereophile in July, some clarification is necessary. I never said, nor do I believe, that "no differences exist between components" (p.47). Such a conclusion is ludicrous and false and a crude misrepresentation of what I have maintained all along (see, for example, S. P. Lipshitz and J. Vanderkooy, "The Great Debate: Subjective Evaluation," J. Audio Eng. Soc., Vol.29 Nos.7&8, pp.482-491).

Far from his claim that the AES excludes "the listening experience from the study of audio" (p.7), this whole paper is about how to conduct meaningful subjective listening tests for the study of audio. In my experiences over a dozen years, audible differences frequently occur—differences which survive even when subjected to a rigorous blind test. These differences, however, have invariably turned out to be due either to a level difference (half a decibel will do it), a frequency-response difference, a difference in polarity, a nonlinear distortion, or an overload phenomenon of some kind in one or another of the components being compared. What I have found, and what I do maintain, is that when these well-known and easily measured performance differences are reduced to sufficiently small amounts so as to be inaudible (a situation which obtains, for example, in most high-quality amplifiers nowadays), the two components no longer sound different.

Surprisingly, and alarmingly, a number of "audiophile" components exhibit frequency-response deviations and nonlinear distortions sufficiently large to be audible, and are clearly wrong in their performance—they alter the sound of their input signals. Whether one "likes" or "prefers" the sound of such a component is beside the point—it is unfaithful to its input, and as such is not accurate. It is amazing how many people eschew tone controls on preamplifiers, but are quite happy to approve of an amplifier, or CD player, or phono cartridge, or loudspeaker which modifies the frequency response in a way which produces a sound which they "prefer" to, or feel to be "better" than, that of a neutral (read "accurate") component.

This brings me to RH's quasi-philosophical discussion (pp.5-15) about what I meant when I asked him how one "know(s) what is good." Unfortunately, he has taken a very simple and straightforward technical point which I was trying to make, and interpreted it to refer to the general philosophical question of how we can ever "know" anything. Please let me explain what I was driving at.

In my view, the better component is the one which more accurately reproduces its input signal. (I totally disagree with JA's value-judgment criterion enunciated on p.144 (footnote 2). If the listener does not like the output of such a component, he should criticize its input, and not blame it for its truthfulness. Such a comparison of output with input provides an absolute basis for assessing the performance of such a component, one which is independent of listener preferences or value judgments—if the output sounds different from the input it is wrong, period.

A comparison of this sort is relatively easily performed on a power amplifier, for example, but is much more difficult when it is a CD playback system which is being assessed, since we do not have direct access to the original signal on the disc. In the quote, "Ah, but how do you know what is good?" (p.8) which RH uses as the inspiration for his essay, we were talking about just such a CD playback system. What I was asking is how he knows whether the player's output signal is a more, rather than less, accurate recreation of the audio signal recorded on the disc. The answer, of course, is that he doesn't know. Again, bear in mind that we're not talking of preferences here—he is at liberty to prefer whatever he wishes—but about which CD player was "better," namely, more accurate.

So, my question actually referred to how one determines what constitutes a "good" CD player. I maintain that there is no doubt whatever that a 3dB rolloff by 20kHz (Stereophile Vol.12 No.9, p.113) is a clearly audible error, whether preferred or not on the particular CDs listened to—or on all CDs, for that matter. It is a built-in high-cut filter masquerading as improved accuracy. I contend that, without a direct reference to the signal encoded on the CD, you have no way of knowing whether what you hear is actually less rather than more accurate. And this is the point I was trying to make in my quoted question.

Finally, I would like to make one more point. It is one thing to mistakenly praise inaccuracy in a component. It is quite another to conclude that an audible difference exists when in fact does not. The former is an error of judgment, the latter one of (conscious or unwitting) self-deception. In my experience, errors of the former kind can be minimized by objective measurements, while the latter kind of errors can be eliminated by properly controlled subjective tests. (Read my paper.) I would be willing to wager that, were I to be present at one of your listening comparison sessions, there would be no unexplained audible differences.—Stanley P. Lipshitz, Audio Research Group, University of Waterloo



Footnote 2: In my review of the Meridian 206 CD player, where I stated: "'better' is when the sound more closely resembles that perceived in real life." I was specifically referring to the ability of "good" components to mimic reality by preserving disparate instrumental or vocal threads within the musical whole, a paradigm I regard as important for good sound reproduction but one more honored in the breach, most components tending to blend individual musical threads into a single synthetic sound.—JA

X