The Chicken and the Egg Page 2

I am sure that in Oberlin Smith's time, it was felt that if only the technical problem of the recording process could be overcome, then everything would be perfect. But even with perfect recording equipment, questions arise as to how it is used, as I endeavored to describe in this space last November. That "As We See It" essay outlined opposing philosophies concerning how to record classical music. First, and one to which I adhere in my own recordings, is to treat the recording process as documentary, where the recordist attempts to capture as faithfully as possible on a two-channel recorder such aspects as instrumental timbre and balance, the ideal relationship between the direct sound of the instruments and their associated reverberation, and the positional information necessary for an accurate stereo image to be reconstructed. (You can see from Barbara Jahn's interview with Riccardo Chailly in this issue how much a world-class conductor likes this philosophy.)

Second, and a philosophy which has dominated the world of classical recording since the early 1960s, is to record the orchestra as though it were a giant rock band. Every instrument or, at best, group of instruments is given its own microphone; the engineer tries to minimize the acoustic leakage from one mic to another; the output of each mic is fed to a separate track on a multitrack recorder; and for the mixdown to stereo, the engineer assigns a precise position along the line joining the two loudspeakers to each track. Artificial reverberation and tonal equalization are often added, to "sweeten" the sound, and, just as with a typical rock recording, the producer "balances" the level of all the tracks to produce what to his or her ears is the "best" sound.

I tried to explain the reasons behind this recording philosophy in Vol.11 No.11, pointing out that its protagonists regard the recording of classical music as a different art form from the live performance of the same music. Fundamentally, however, I have to say that such a stripping down of a musical event to its bare essentials at the time of the sessions and attempting to put it back together again during the mix is doomed to failure. Think of the Second Law of Thermodynamics—think of Humpty Dumpty!

An interesting letter this month from Hank R. Bernstein comments on one of the conclusions I drew in last November's essay: that it is possible to judge the quality of a hi-fi component using music that has no original reference; rock or jazz using electric instruments, for example. Mr. Bernstein feels very strongly that this is fundamentally unsound in that the reviewer is then unable to communicate any real description of the component's sound to a reader: "In the end all that reviewer can tell me is that he liked or disliked what he heard: that the sound of the electric guitar seemed to come from a proper position on the stage, that the effect was quite dramatic as the sound was made to cascade across the listening stage at some propitious musical moment, that the component under consideration reproduced these engineered effects well, or better than brand X component which he had reviewed or listened to the month before."

Without the absolute reference standard of unamplified acoustic music in its performance venue, a standard suggested by J. Gordon Holt in last October's "As We See It," Mr. Bernstein suggests that the reviewer's conclusions are useless to the reader. He goes on to say that, with the absolute sound used as a reference, however, the reviewer can offer the reader observations about reproduced sound that are meaningful: "He may tell me that the music is recessed, as if I were in Row R instead of Row D; that the timbre of the cellos is true but that the sound of massed strings is a bit more brittle than what one would hear in live performance."

In broad principle, I would not take issue with any of Mr. Bernstein's reasoning. However, I have to point out from a practical standpoint that a reviewer who assesses components solely on whether they help a system more or less closely resemble the sound of live, unamplified music will paradoxically very quickly fall into error! This, of course, is not because the absolute sound of live music is not something to which we all aspire, but because it is almost never available in the context of recorded music except in very broad background terms.

If this sounds confusing, let me examine the implications of the statement that it is more useful to the reader to be offered the observation that a component makes a classical recording sound as if the listener were "in Row R instead of Row D" at a live performance. As a piece of anecdotal information, describing what happened when a component was inserted in a "reference" system, this statement can't be criticized. However, as a value judgment it falls short of the mark, for it assumes that the record used to form the observation has within its grooves the ability for a perfect system to reproduce the sound of live music in the listener's room.

'Tain't so, fella. Only if a classical record has been made with everyone involved—from musicians to recording engineer to producer to cutting engineer—concerned with the accurate preservation of every aspect of the original sound and performance, will this assumption be correct. Otherwise, the listener has to use a recording that is no different in kind from a multitrack rock recording to judge the sound of a component, and that record will have probably been considerably altered from the "sound of live unamplified music" in just about every way.

Attempting to use the absolute sound—love that phrase!—as a reference, you may think that you can make value judgments about a hi-fi component because you already know what a classical recording should sound like—the real thing. In actuality, as the recording doesn't have that potential, you need first to assess its quality by listening to it through components that you need already to know the sound of when compared with the sound of live music. You need knowledge of the chicken before you can assess the quality of the egg, but you only have access to the quality of the chicken if first you know all about the egg.

Confused? Consider the statement that a component makes the sound "recessed" when compared with the real thing. There are two, not one, value judgments that can be drawn: the first is Mr. Bernstein's: ie, that the component makes the sound more recessed; the second is that the component more accurately allows the intrinsically more recessed sound within the grooves of the LP to be reproduced. Which is correct? You have no way of knowing.

Read Robert Levine's review, for example, of the new Michael Tilson Thomas recording of Mahler's Symphony 3 on CBS Masterworks in this issue, where he notes that it "sounds as if each solo instrumentalist...had a separate microphone on his lap," resulting in "the most crisp, spotless Third imaginable." I suggest that a reviewer using this recording to assess the quality of components would downgrade neutral-sounding components on the grounds that they made the sound less like the sound of live music; ie, it would sound too bright. Components that are intrinsically too dull will be unjustifiably upgraded for the opposite reason.

This causal dichotomy holds true for every aspect of sound reproduction, not just tonal balance. (Visitors to either of Stereophile's 1987 hi-fi shows will remember that I devoted a whole hour's lecture to just how a recording's ability to create a stereo soundstage is altered by the microphone technique used, footnote 1.) The conclusion must be obvious: use only those classical records in listening tests that have been made according to philosophy #1 above. Use program material that is intrinsically neutral in tonal balance, that has the instrumental balance that the composer intended, that is uncolored, and that has the capability to create a true stereo image.

OK, name some records that you know conform to this standard.

Name one.

Difficult, isn't it? In fact, I can only think of one commercial orchestral recording that gets close to this ideal, Sheffield Lab's Firebird, with Erich Leinsdorf conducting the LA Phil (CD 24), and even that is flawed: it was recorded in what I feel was an unsuitable acoustic, too dry and with too small a volume for such a dynamic work. In fact, you have to realize that recording engineers, even when they aim for honesty in recording, are always forced to compromise some areas of reproduction in favor of others. Les Berkley, for example, reviews a new Hungaroton production of Alessandro Scarlatti's oratorio Judith this month, where the engineers chose to use microphones that were timbrally accurate. However, they also used a mic technique, probably dictated by the very natures of those microphones, that destroyed any sense of a real soundstage.

Foonote 1: I was intrigued to read in the November/December 1988 issue of The Absolute Sound that pianist and educator James Boyk is offering a commercial cassette, realized by some of his students at Caltech, which appears to be based on my series of lectures on how different stereo microphone techniques affect the reproduced soundstage. Readers interested in acquiring this tape should contact Mr. Boyk at Performance Recordings, 2135 Holmby Avenue, Los Angeles, CA 90025-5915. Tel: (213) 475-8261.—John Atkinson