An essay on Neutrality: Fallacies and other issues

Roy Harris

What is neutrality?

Denotations include: unbiased, impartial, fair minded, belonging to neither of two classes,  a middle position between two extremes and having no color, e.g., water.

In audio terms neutrality is an absence of coloration, i.e., having no affect upon the sound of the recording. If a stereo system were neutral one would expect to hear the sound of the recording, as is, no more, no less.

Unfortunately, it is easier to comprehend the meaning of neutrality than it is to identify neutral components and stereo systems.

Are stereo systems or components neutral?

Since components are imperfect, neutrality is an ideal, which has not and may never be attained.

If distortion is detected, a stereo system cannot be neutral. However, what happens when components have levels of distortion below the threshold of audibility?

It is possible that stereo systems configured with components of this type may be deemed neutral if a sonic signature is not detected.  The fact that coloration is not (subjectively) detected does not mean that there isn't any. It does mean that perception is not sufficiently discriminatory, i.e., one's JND is too large. The JND is the magnitude of the difference between two stimuli necessary for distinguishing one from the other. One could use a more appropriate term to describe such stereo systems,            namely ‘virtually’ neutral. This designation may be subject to disagreement among serious listeners. One would not expect a consensus of opinion      regarding judgments of this nature.

The implementation of the concept of ‘virtual’ neutrality has as its beneficiaries, audio professionals, especially reviewers and manufacturers.

Reviewers usually criticize the presence of coloration. Manufacturers who assert neutrality as their design goal may be motivated by their desire to avoid criticism in the press. Other reasons include the belief that neutrality represents the pinnacle in design, and a marketing strategy that assumes consumers demand products to be as minimally colored as possible at a given price point.

The fallacies discussed below are based upon the conventional wisdom of the high-end audio industry.

Consider the statements:

     (1) Components and stereo systems sound different from each other.

     (2) Some components and stereo systems are neutral.

Statement (1) represents the attitudes of many audio professionals and serious hobbyists.

It is counter productive for an audio dealer to assert that all components sound the same. If reviewers allege that all amps, preamps, etc., sound the same only one from each category need be reviewed and only a few reviewers would be necessary to review them. Thereafter, reviewers might be superfluous.

If equipment purchasers believed that all components of each type sound the same, the number of manufacturers able to sustain a viable operation would be very small. Thus, it is a clever marketing/sales strategy to promote the idea that components and stereo systems sound different from each other, whether in fact they do and whether those who maintain this position believe it    themselves.

However, a problem occurs when one insists that components and stereo systems are neutral, i.e., when statements (1) and (2) are accepted as fact. Statements (1) and (2) are contradictory, creating the first of two fallacies.

While it is in the interests of audio professionals to try to persuade consumers that the products they design and sell are neutral (accurate), they expose themselves to criticism for espousing two mutually exclusive propositions, namely that components and stereo systems sound different      and are neutral. It follows from statement (1), that components and stereo systems have a ‘sound’. It follows from equation (2) that components and stereo systems that are neutral have no sound. How can a stereo system sound different from another and yet have no sound? Conversely, how can stereo systems, which have no ‘sound’, ‘sound’ different from each other?

I believe what manufacturers mean to say is:

        ‘My products are designed to be as accurate as possible at a price point’.

The above statement implies that a component is not neutral and can sound different from another.

The above premise may place some dealers in an untenable position. It is more efficient for a dealer to recommend a product based upon the belief of neutrality. It may be an easy sell as the need for audition may be eliminated. It is also possible that dealers do not want to admit that their wares are colored but also want to give the impression that the products they sell sound different, thereby providing a choice to potential buyers whose preferences may differ. Thus a dealer may try to cater to a customer's sonic palette while maintaining that his products are accurate or almost accurate. The dealer is on logically shaky grounds if he upholds the belief that components sound different and are uncolored. On the other hand, he may mislead his customers when insisting that components are minimally colored and suggesting the coloration is so slight as to be unnoticeable.

A variant of the above mentioned fallacy arises when, in lieu of statement (1), the following is alleged:

     (3) Some components and stereo systems are better than others

Asserting statements (2) above and (3) creates a non sequitur. Statement (3) implies that components have a sound. How can components be neutral if they have a ‘sound’? If components have a sound, hence are not neutral, on what basis is one better than the other?  A typical answer would be one is less colored than the other. What is the probability that two components will be in a perfectly linear relationship with respect to all forms of coloration? Isn't it more likely that the quality and quantity of inaccuracy will differ among two or more components? If this is the case there is no logical grounds to assert one component's superiority   over another when both components essentially are inaccurate and have a different sound.

Unfortunately, it seems reasonable to accept the idea that components vary in sound quality. Such an approach is an obvious marketing tactic for manufacturers and dealers as it supports price differentials for components. However the concept of sound quality does not rest on logical

grounds even though one may try to construct an objective basis for it. One would have to demonstrate that a component has lower levels of distortion, in all its manifestations, to assert its superiority over another.  Such a task is a challenge as there may exist types of distortion, which are not yet measurable or identifiable, but yet audible.

Finally, here's the paradox:

Suppose there exist two stereo systems, I and II, which are judged (independently) ‘virtually’ neutral. It is likely that when I and II are compared directly differences between them will be perceived, especially if the speakers from each system vary in design characteristics, e.g., panel and cone. There are two paradoxes to consider.  First how does one reconcile the paradox that two stereo systems evaluated independently are deemed ‘colorless’, but when reevaluated in an A-B comparison mode, differences are detected such that the ‘alleged’ ‘virtual’ neutrality of each is nullified?  Secondly, if a stereo system is judged ‘virtually’ neutral, how can it have a ‘sound’?

Both enigmas can be explained using the principles of psychophysics, a quantitatively oriented branch of psychology.

In its simplest terms, judgments and perceptions are more accurate when there is a reference. When comparing stereo system I and II, one is the reference, control or anchor, relative to the other. The process of evaluating I and II separately and independently is more complex than describing the differences between I and II. Comparative judgments are easier to make than absolute judgments.

The contradiction described above is rooted in a listener's inability to perceive subtle cues of coloration under certain conditions. It calls into question the validity of evaluating stereo systems. Subjective audition without a reference increases the likelihood of errors. The results of induction, the way in which stereo systems are usually evaluated are inconclusive, probabilistic and uncertain. After listening to ‘n’ CDs, one cannot be sure that one's (initial) judgment won't be reversed when listening to one more.

Given differences in speaker design and their high levels of distortion, it is unrealistic to expect any two stereo systems with different speaker systems to sound the same. Further, one would expect a very experienced listener to notice some sonic signature when evaluating a stereo system, in the absence of a reference, as it is easier to detect a flaw than be sure that none exists.

Conclusions

In the ‘universe’ of stereo systems and components, the following are possible:

      1) All are colored

     2) Some are colored

      3) None are colored

Based upon the fact that there are no perfect components, one must accept statement 1) above. Here are the implications:

      A) Some components have audible levels of coloration

      B) Some components have levels of coloration, which are inaudible

Proposing that coloration may be inaudible raises several issues, namely the sonic evidence necessary to assert the absence of observable coloration and the assessment methodology.

Without a reference or benchmark, the typical method of evaluating stereo systems and components is likely to yield unreliable and erroneous results, as it is easier to identify what you hear than what you don't hear. In addition, it is likely that some ‘expert’ will not concur with the opinion of others and claim that coloration is present. The resolution of such a dispute can never be definitive. At best the finding that coloration is undetectable is a hypothesis that is untestable and can't be proven.

An objective measure of coloration is needed. Since there is more than one type of distortion it is advisable to specify each and measure it. When meaningful quantitative estimates of distortion are available, it is useful to correlate them to listeners' perceptions of components and stereo   systems. Studies could then be published giving prospective purchasers more information than they have now.

Currently there is a way to obtain an indication of the (in)accuracy of interconnect cables, speaker cables and line stages. David Salz of Wireworld has used this simple approach I am about to describe.

It is a bypass test. The requirements include a digital source, amplifier and a pair of speakers.     Hardwire all three, listen and measure.  Introduce an interconnect, listen and measure. The difference between the sound of the stereo system with and without the interconnect should be  noted. This process can continue as a line stage and speaker cable is added. Two or more cables, etc., can be compared using this paradigm. Under what would be construed as control conditions, one would obtain ample evidence that the most carefully designed components have levels of distortion that, although slight, are noticeable. This fact might suggest that stereo systems and components have audible levels of distortion, but that under ‘normal’ conditions and in some cases, listeners fail to notice them.