Who's Right? Accuracy or Musicality

Many years ago, the now-defunct Life magazine ran a feature article about science and its sacred cows, in which a cartoon showed a huge inverted pyramid-shaped structure of great complexity, tapering downward to a single support at its base: a toothpick. The toothpick was labeled "basic premise," the inverted pyramid was the entire body of scientific knowledge.

Everything we do or think or know is based upon assumptions, some of which are rather more justified than others. When we set the alarm clock, we assume there will be a tomorrow. When we reach for the car's brake pedal without glancing at it, we assume it will be where it was yesterday, and that it will stop the car. When we scorn a phono cartridge because it is too bright, we assume the brightness is in the cartridge, not in the rest of our system. We have to trust our toothpicks or live in a world totally devoid of security—a world where 2+2 can equal anything from 3 to 11, all the laws change unannounced every few days, and Greenwich Mean Time is determined by a roulette wheel.

Most audiophiles are insecure. Some of us distrust our own judgement of what does and does not sound "good." Those of us who do have confidence in our judgement worry because, somewhere out there in audioland, there may be a better preamp or cartridge or loudspeaker than any we have been able to audition locally. So we all read the "underground" audio press, hoping to be told what is the best of each kind of component and what we will have to put up with in terms of deficiencies if we buy something less than state of the art. And we are making an assumption: Namely, that the people who write the reports for those magazines are good judges of equipment themselves. We must trust the toothpick.

The problem is, the magazines themselves have made it impossible to trust that toothpick, each by building its own inverted pyramid—its "credibility"—on a different toothpick. Magazine A top-rates preamp B, magazine C downratcs that preamp and chooses preamp D, magazine E finds prcamps B and D to be gritty-sounding and favors preamp F. And not surprisingly, the reader becomes a little bewildered. "How," he asks himself, "can they disagree so much, not only over what they like, but what they hear?"

The reason of course is because each of those magazines, too, has made a basic assumption, and that is that their standards for judging components are Intrinsically Right.

Since fidelity is the name of the game, accuracy is of course its objective. But before an equipment reviewer can determine in what ways a component is inaccurate, he must first determine what is the "right" sound. There must in other words be some starting point—something that he can assume to be right, and on which he can base all his subsequent judgements about all of the components he tests. Once he has made that assumption, it is necessary for his peace of mind (and to give him the courage of his convictions) that he not harbor any lurking doubt that his basic assumption—his supporting toothpick—might in fact be incorrect.

For it is exceedingly difficult to prove the intrinsic rightness of anything subjective in audio—so much so that it can be said with confidence that no component reviewer can prove the correctness of his basic assumptions through irrefutable logic. That is why an equipment reviewer gets so testy when his observations are challenged, for that challenge is a profound threat to his personal security. What it says, in effect, is: "Your observations about that component are wrong because your standard of reference is wrong, and if your standard is wrong then you have made wrong observations about everything you have ever tested." That kind of thought is something a reviewer can live without!

Suppose the tester does inadvertently build his credibility on an incorrect basic assumption: The assumption, for example, that a certain recorded bass/treble balance is correct when in fact it is thin and bright. Wouldn't his readers soon find this out and cease to trust his reports? Not necessarily, for once a reader has followed a couple of his recommendations, that reader's system will start to take on the same set of sonic "biases" as the tester's reference system. From that point on, said reader will find, more and more, that his own observations about the components he buys agree with that tester's assessments of them. To the reader, this is vindication of his confidence in that tester, but the reader will have unwittingly become just as effectively trapped by that closed circle of incorrect assumptions as is his favorite tester (or magazine).

It is often said that the reason people disagree about loudspeaker systems is because "different people hear differently." This is usually interpreted as meaning that their hearing equipment differs, but that is not the idea at all. Everyone's ears respond to sound waves in the air, and if those waves in the living-room air are exactly the same as in the concert hall air, everyone will hear exactly the same thing in the living room as they would in the concert hall. Each may be hearing something different from all the other listeners, but if the sounds reaching his ears are an exact replica of the original sounds, he will interpret what he hears as a literally accurate reproduction.

The differences of opinion about reproduced sound arise because people differ as to what aspects of the sound—original or reproduced—that they listen for when making evaluations, and what relative importance they assign to each such aspect. Equipment reviewers for underground publications are like other mortals in that they tend, consciously or otherwise, to "weight" their evaluation of sound reproduction according to those aspects of reproduced sound of which they are most critical. Thus, even while describing in seemingly impartial detail what a component is doing to the sound passing through it, they (and we must include ourselves, on past record) rarely feel compelled" to mention the fact that those listening tests were conducted in conjunction with other components that they previously chose as "best" for testing purposes according to their own personal weighting of aspect priorities.

Stereophile for example feels that, at a time in audio history when superb bass and treble and excellent detail are obtainable from many loudspeakers, the most important remaining consideration is freedom from coloration through the middle range (where, after all, the vast majority of musical activity takes place). We have continued to use and to recommend Fulton Musical Industries J speakers because, despite frequent modifications and manifest weaknesses (ie, low-end and and midrange detail easily bettered by some other systems), they offered what we feel to he the best set of performance compromises of any available speaker system. The FMI Js, in other words, while not quite the equal of some others in some respects, are so markedly superior in their reproduction of real musical timbres that they must be considered one of the most accurate reproducers of music around today.

Our feelings here have been confirmed numerous times by visiting professional (classical) instrumentalists and conductors who, while acknowledging that the Dahlquists and the Magneplanars were more exciting to listen to, have opined that the Js make the instruments they are most familiar with sound most natural to them. Yet one other "underground" audio magazine has explained that it uses the Dahlquist DQ-10 as its reference system because of its "accuracy," and rejected the FMI Js because they "glorify" or "prettify" the signals fed to them. And another magazine, while giving high marks to the DQ-10, has continued to use Magneplanars as its reference system because of their "accuracy."

Just what, then, is it that constitutes this musical accuracy that every audio magazine adulates hut few can agree about? Is it detail? Certainly that is part of it, but it does not seem to have occurred to some of the equipment reviewers that detail can be exaggerated, so that the system sounds as if it is reproducing more of it than is in the original program material. Boosting the 7.6kHz level on an octave equalizer can improve the apparent detail in the sound. So can certain kinds of distortion, which are less obvious because they cause no measurable change in frequency response. A speaker system with a "hot" high end will do the same thing. Is it the more accurate because it reproduces microphone peaks and disc mistracking more conspicuously than does another system? The answer is: Not necessarily.

One editor said, editorially, that top-line audio seems polarized these days into two "factions": those purveying "accuracy," as exemplified by the Mark Levinsons and Dahlquists, and those purveying "musicality," typified by GAS electronics and the Fulton J speakers. His magazine, he proclaimed, was dedicated to accuracy. From the way things have been going in audio publishing lately, we figured it was just a matter of time before some editor would declare that fidelity (or accuracy) and musicality are two different things. But this one then proceeded to ram his other foot in his mouth by allowing as how musicians are more "comfortable" with the musical kind of reproduction he had just officially renounced.

It is true that most classical musicians are lousy judges of most aspects of fidelity; they listen to the playing, not the sound. But it has been our observation that a musician who spent $20,000 for an instrument that sounds, live, exactly the way he wants it, gets very uncomfortable indeed when it comes back at him sounding like something out of the high-school orchestra instrument locker.

Of course accuracy is implicit in the whole idea of fidelity, but accuracy according to whom? Again, it is a matter of what the individual listener is most critical of, but it seems quite apparent that audiophiles and most of what we call musically oriented people do indeed listen for, and value, quite different things.

The editor who chose accuracy over musicality admitted that there was nothing inherently reprehensible in preferring a system that pretties up the sound, and we agree. But let's assume, with him, that if a system is truly accurate it will sound musically natural to anyone—audiophile or musician. We like to believe that what we are all striving for is a system that will reproduce exactly what is on the recording, be that good, bad or so-so. Such a perfect component should perform its desired function—transduction, amplification or what have you—and nothing more. It should neither subtract from nor add to the input signal. It should, in other words, be the equivalent of a straight wire with gain, and when inserted into the signal path, it should produce exactly the same sound as when it is bypassed. Nothing could be more self-evident.

It is possible to bypass-test a preamplifier or a tape recorder. Anyone who does this, and properly interprets his findings, should be able to do a definitive report on either one of those products. Interestingly, reviewers sometimes disagree about these, too, but that's beside the point. The point is that there is no such way of bypass-testing any other component in the entire audio system!

Different recordings sound different in a number of ways for a multitude of reasons. There is no agreement within the industry us to which recordings are correct even in bass/treble balance. Even the best professional microphones differ in sound (fig.1). All loudspeakers differ, and commercial recordings are equalized to sound "right" on whatever monitor speakers the studio selected (on the basis of sounding "right" with the microphones they selected). Disc-cutting systems sound different, too.

Fig.1 Manufacturers' published frequency response curves for two professional cardioid condenser microphones: the German Sennheiser MKH-405 (above) and the Japanese Sony C-37A.

How, then, do you tell what a disc is supposed to sound like? And if you can't, then how can you tell whether a cartridge is making it sound the way it should? You can't listen to a cartridge anyway without using a preamplifier which, while assessable via a bypass test, may nonetheless interface poorly with that particular cartridge. Amateur tape recordings are a more reliable program source than commercial recordings, but again, what do those microphones really sound like? Some are brighter than others, but which are "right"? As for power amplifiers and loudspeakers, these are so interrelated in function that it is almost futile to try evaluating the sound of one without knowing exactly what the other is doing (fig.2). In fact, in a better world there might be no separate amps or speakers, there would only be integrated amplifier/speaker systems.

Fig.2 An example of complementary imperfections. Frequency-response curve A of the EMT phono cartridge, complements curve B of the Magneplanar 1-C speaker system, to produce almost-perfectly flat net response of curve C.

There, in a nutshell, is why there are almost as many opinions about reproduced sound as there are "underground" magazines. Each claims exclusive proprietorship of The Ultimate Truth when the reality of the matter is that there is no Ultimate Truth in audio. There are only value judgements. And what is it that makes one audio critic's judgement better than anothers? A solid technical background plus plain, old-fashioned experience.

For example, it helps to have lived through the transition period from 78-rpm discs to the LP, and to recall that the major shortcomings of recordings then were a result of poor recording equipment rather than excessively "creative" producers. As long as the microphones and disc-cutting equipment were manifestly imperfect, it was considered a challenge to be able to produce the highest-possible-fidelity disc with the limited means on hand. Since those late-1940s recording engineers were aware of how imperfect were their loudspeakers, too, they relied more on frequency-response measurements than on what they heard when equalizing their microphones and cutterheads. The only other equalization that was used (generally) on the disc was the standard (within each company, that is) bass cut and treble boost which were supposed to be compensated for in playback (but rarely were). Interestingly, as their equipment improved, properly-equalized latter-day 78s, and then the first LP discs, provided increasingly similar overall bass/treble balance on playback. There was less variation in that sonic parameter then than there is now!

Today, there is still much to be learned from a careful perusal of the frequency-response curves published by manufacturers of professional recording microphones. These, we have found, are nearly always accurate, sometimes to the point of embarrassment. The typical German cardioid condenser mike's response curve, while almost ruler-flat over most of its range, shows a gradual upward tilt from low to high end: Clearly the source of that lean, coldly analytical sound. The Sony professional cardioid condensers, on the other hand, are horizontal across most of the range of their published response. And what do you know: Their overall balance sounds like most latter-day 78-rpm discs and the early LPs. (It goes without saying that the Sonys are vastly better in all other respects.) We aren't saying that that proves the Sony mikes are right and the brighter-sounding ones are wrong, though. There are no absolutes here, remember?

But there are many other clues as to the relative rightness or wrongness of certain kinds of sound. The showy brilliance of some early amplifiers, which helped to launch the Acoustic Research loudspeaker because the latter helped soften that raw edge, vanished in later amplifiers as distortion figures were reduced to well below 0.5%. Old loudspeakers that had been judged as harsh on those bright amplifiers were vindicated. Some are still in use by recording studios. Cartridges, like microphones, tend to sound the way their frequency response looks as if they ought to, if we can accept the fact that a 1dB response deviation over a fairly broad band is quite definitely audible. Most samples of the Denon cartridges have a brightness suckout of that amplitude, yet it is one of the most popular perfectionist cartridges. If the resulting sound is "musically accurate" on a trustworthy recording (?!), what docs that say for the rest of the system it is being used in?

Then there are the consumer reactions. If most owners of a certain component find it to have the same strengths and weaknesses, that is an indication, although certainly not proof, that it is indeed that way. Who likes it and who dislikes it? Audiophiles and musical people tend to like different things, and over a period of many years we have seen what kinds of sound are preferred by each and, more important, what kind is very often preferred by both.

We used "kind" in the singular there because, in the 15 years we have been publishing Stereophile, we have seen emerge a pattern which delineates a middle ground and a rather definite pair of limits to what constitutes "tightness" in reproduced sound. That, plus 12 years of experience recording live music from chamber groups to orchestras, is what gives us a little more confidence in our basic assumptions than we might otherwise be able to muster.

Of course, we're obviously biased. And you couldn't really expect us to conclude this dissertation with a confession that we don't know which end is up. But let's put it to you this way: If you had to have some touchy surgery where judgement and skill mattered, would you choose an intern to do it, or an old pro? Think about that when next you're looking for a speaker system.

Rvpiano's picture

Alleluia for this article!
You can no more say that there is a definitive high end system than you can say that Bach or Mozart or Beethoven is absolutely the greatest of all composers.
There are standards, but not absolute ones. Musicality (in the ear of the concert going listener) trumps accuracy (whatever that is!)