Music Before Numbers? Letters

Letters on this subject appeared in the March and June 2001 issues:

Those measurements

Editor: I must applaud Stereophile for its determined efforts in including objective measurements in its coverage of high-end products. Although measurements often cannot discern for us the differences we do hear (I believe in mostly subjective evaluations), there comes a time when I, at least, must decide to purchase a component as a musical instrument, or one that attempts to reproduce a neutral representation of the software presented it.

The previous controversy over the "tailored" frequency response of some highly reviewed loudspeakers and the present spate of tubed components with ghastly measurements shows that the musical-instrument fraternity of equipment manufacturers is alive and well.

This is not an attempt to suggest which option you should prefer, but without Stereophile's attention to measurements, how would we even know that this exists? Keep up the excellent work.—Dale W. Smith,

Those bench-test results

Editor: I like the fact that Stereophile includes bench-test results along with the listening tests of the components that are reviewed. I know, of course, that standard bench-testing might not tell the whole story. I have a solid-state preamp from the late 1970s that has excellent specs—vanishingly low distortion, etc. But my Dynaco PAS still sounds better to me. (The Dynaco also has high-fidelity credentials, though not as impressive as the solid-state unit.) But I believe having both is the best way to get a fair idea of the worth of the equipment.

This letter is not meant to disparage the Cary power amplifier (which I have not seen or heard) reviewed in the December issue; it appears to be a well-built and attractive piece of equipment, and should appeal to those with the means to indulge it. However, I do take John Atkinson's point (as I understand it) that, when listening to equipment aimed at the high-end audio market, we might reasonably expect it to meet current high-fidelity standards when tested.—Craig D. Merz,,

You have to listen

Editor: I have gotten up to write this letter many times in the last few months, but have always put it off till another day. I guess, with the subject of April's "As We See It," that day has finally come. This debate—over whether a component's measured results have anything to do with what you hear from that component—is a prickly one because no one can know how well another person hears, or how well the measurements were performed. Nevertheless, I will attempt to address this issue evenhandedly.

You can argue, as objectivists are wont to do, that two preamplifiers of identical specifications will be interchangeable. If you believe this, then stop reading now; you will only be outraged at my irrationality. Or you can argue, as do many subjectivists, that there is no sense in reducing noise and distortion in components, that instead you should focus on getting in touch with your inner audiophile. If this is your philosophy, you may want to stop now before I destroy the "magic of the music" for you.

For the two of you still reading, my stance is that while measurements are an excellent indication of how consistently a component will perform with a wide variety of media and software, only by listening can we discover the limitations of the component within the framework of the measured results.

An illustration: A single-ended triode amplifier will perform admirably with high-sensitivity, high-impedance speakers provided you don't push it too hard and expect it produce low frequencies at high SPLs (timpani, pipe organ, kick drum...). Its measured performance indicates this, but to find specific limitations you must listen to a given model and find out whether or not this applies. Additionally, an amplifier topology as quixotic as this will never allow you access to improvements in the resolution of your front-end. You may be able to hear that you have changed something, but a high-resolution, dynamic, low-distortion signal passing through a SET amp will no longer be any of those things when it comes out.

Conversely, if your speakers present a benign load, and if your shelves are full of chamber music and small-ensemble jazz, then to your ears a SET amplifier may be the best thing available. The measurements indicate this as well. But if you favor monstrous speakers and a varied diet of music, then a SET will quickly wear out its welcome because its distinctive sound will be coming between you and your music instead of adding to your musical enjoyment. This is true of any deviation from neutrality, not just "the SET sound."

I thereby propose a correlation between a device's total deviation from neutrality and, not its sound, but its compatibility with a broad number of other components. The exceptions to this rule are bleeding-edge designs that reveal other devices' deviations due to their own lack thereof.

Of course, the only way to determine compatibility is to listen to different combinations of equipment and find one you like.—Michael D. Odom,

You have to measure

Editor: There's an old saw: "If it measures bad, it is bad. If it measures good, it may not be good." Like all problems in which no one has the whole answer, everyone has a bit of the answer, and the best answer comes when we all cooperate. We need to intelligently listen and measure.

Stereophile was founded on the idea that how a component sounds matters most (I have read every issue since the first). But it is important to understand that this doesn't always mean a product is accurate because it sounds "good" or "musical." Indeed, if it always sounds "good," it isn't accurate. All software isn't accurate, and accurate equipment will sound bad on lousy software. How we recognize accurate software is another problem, since we listen to software on the equipment we are testing.

But it would be good if measurements could tell us more of the story. Measurements allow repeatability and predictability. Both of these are good for designing technical devices such as audio gear. And I must admit a bias, perhaps: When a device measures very badly, I tend to discount its accuracy no matter how it sounds.

As an example, and at the risk of incurring the wrath of single-ended amplifier lovers, I intellectually don't like single-ended amps. They measure awfully. Indeed, into a reactive load, as every loudspeaker is, they are incapable of reproducing a simple sinewave; they compress the negative half of the sinewave. How, then, can they possibly accurately reproduce the much more complex waveforms of music? They sound "good" because this nonlinearity leads to their high-second-order harmonic distortion—a very "musical" distortion that is pretty, not accurate. My Western belief in cause and effect finds this intellectually intolerable.—Allen Edelstein,