Down With Flat!:
AHC Responds part 2 These problems also interact in terms of perception. The listener cannot separate "flat" response in the treble, midrange, and bass. A speaker must. have coherent balance, which is extraordinarily difficult to measure using frequency response tests. For example, most small monitors with flat extended highs and an early rolloff in the bass sound very bright, even though they measure flat within their frequency bandwidth. A classic case is the ProAc Tablette, which I, at least (footnote 2), regard as an excruciatingly bad speaker because the high frequencies dominate over a complete lack of bass. Other speakers (like the Boston Acoustics), however, sound "heavy" because they roll off the upper midrange and highs. How extended should "flat" be in the treble and bass to be heard as "flat?" What balance of treble and bass energy is right? You can hear it, but no one agrees on how to measure it.

The net effect of all of this is that any reviewer who bothers to test speakers and cartridges in terms of frequency tones is going to give up their membership the "flat earth society" early in their career. Speaking personally, I have liked speakers, cartridges, and other equipment with radically different upper-octave energy levels, so has every published reviewer I know of. The issue is not whether a given piece of equipment measures "flat," but whether it is musically convincing. Many different solutions to upper octave energy sound convincing.

Flattening The Music:
And remember this: even if we could solve the steady-tone problems of speaker measurement, music remains an extremely complex transient waveform with an incredible number of different pulses and dynamics. No current method of testing—FFT, pulse response, etc.—approaches the abilities of the human ear in assessing the dynamic character of frequency response. Electronics exhibit significant audible differences in apparent speed and detail in the upper octaves, even when equalized to show identical response with pink or white noise.

Cartridges and speakers exhibit far more of these differences, and, in terms of perception, to far greater effect. The moving-coil versus moving-magnet controversy is a good case in point. Many people honestly cannot hear the difference between an equalized Shure V-15 and an Argent Diamond. Others not only hear the difference, but regard it as absolutely critical. The same is true of electrostatics versus ribbons versus cone speakers versus EMITs, Dolby and dbx, tonearm resonance, and the sound from CD players.

Even when all this equipment is equalized to measure "flat" using the same test tones at a given listening position, all still reproduce the upper octaves in audibly different ways. This is at least partly the result of the fact that each driver has a different way of reproducing upper octave pulses, but the full explanation is far from clear.

The Flatter Ear:
All of this not only makes technical measurements of frequency response questionable, but also creates a problem in subjective reviewing that all of us would like to ignore. The extreme complexity of the ways in which a given device can treat upper octave musical information means that reviewers almost inevitably do not agree on their judgments of what is right, nor in their descriptions of the differences between products.

This is particularly true because it is impossible to review the upper octaves in any given audio component without introducing at least six different components (signal source, turntable or tape unit, preamplifier (usually), amplifier, speaker, and cables). Each element alters the fine dynamics of music in slightly different ways, and the chance of any two reviewers listening to the same chain of components is almost zero.

Now, God unquestionably will bless the one reviewer who proves to be right in his judgment of "flat" on the day of the Last Judgment, and God will spend a few thousand years burning the pointed, golden ears off the rest. I keenly look forward to such vindication.

In the interim, however, it's necessary to point out to the readers of Stereophile, who are struggling to determine what really is flat in the perceptions of various reviewers, that I generally prefer moving-coil cartridges while JGH prefers moving-magnets, that I like ribbon speakers while JGH yearns for the electrostatic high end, and that I am generally more critical of CD players than is Gordon (sometimes radically more critical). This kind of difference doesn't make me right and Gordon wrong. It represents a legitimate difference in what we hear, and demonstrates the need for, and the difficulty of, very complex value judgments regarding different kinds of upper octave performance.

Further, it should be obvious to any literate audiophile that the value judgments of Harry Pearson, Peter Moncrieff, Len Feldman, Theodore Beaufort, and Julian Hirsch differ broadly from those of both Gordon and myself. Even were we to agree on terminology, our differences over what upper-octave performance is correct are unlikely to go away. Nor will some set of the latest computer gimmicks help prove which reviewer is revealing ultimate truth.

I do happen to believe that I am flatter in my judgments than Gordon, they (the other reviewers), or thou (the reader). But, I may be flattering myself. As Abraham Lincoln once said, all men are not created equalized, One can fliply flounder in flatter, and do nothing but flightly flitter to flutter (or wow).

This discussion of "flat" may help audiophiles understand why they hear radically different things from equipment than they were led to expect from reviews, why reviewers disagree, and why equipment that measures well using comparatively simple tests sounds bad when it is used to reproduce music. It may even help to explain why reviewers keep harping on the need for readers to rely on their personal judgment and listen for themselves.—Anthony H. Cordesman

JGH Replies:
After having read several times AHC's erudite dissertation on frequency response, I am still not sure whether he agrees with me or disagrees. The whole point I was trying to make—vainly, it would appear—was that the "flat frequency response" should not be accepted as the last word in the design of transducers in general and loudspeakers in particular. Part of the reason for this, as AHC has pointed out above, is the great difficulty of obtaining valid frequency response measurements. The rest of the reason is that—regardless of the way measurements are made—loudspeakers persist in not sounding the way their measured response would indicate that they should. Even when heard from a close seat, live instruments never have as much HF energy—harmonics, attacks—as does a loudspeaker whose high end measures flat at the listeni~ng seat.

Whether that measurement is taken on-axis or off- is irrelevant. It's what reaches the listening area that counts. If a speaker can image satisfactorily without being toed-in, then the measurements will be off-axis. If toeing-in is necessary, an on-axis measurement will be more accurate.

Similarly, loudspeakers which measure flat through the middle range tend to vary all over the place in their apparent midrange output. Some sound very laidback, others sound quite forward.

We all agree that a trained pair of ears is the ultimate arbiter of sound quality. (That, after all, is the sole justification for Stereophile's existence.) I am merely suggesting that loudspeaker designers bear this in mind the next time they offer a flat measured response as proof of excellence. It ain't.—J. Gordon Holt

---

Footnote 2: AHC's differences with JGH may seem extreme, but not on this speaker. He didn't comment on Dick Olsher's report in Vol.7 No.4, but his reaction to the ProAc's frequency balance was "appalling."—Larry Archibald