The Fifth Element #5

"The long tradition of professional connoisseurship has resulted in the development of a bewildering universe of specialist terminology. In certain cases, it must be admitted, there was self-indulgent proliferation of words relating to some minute feature....In fact, no clear distinction can be made between one term and its closest neighbor in meaning."—from the Introduction to Kanzan Sato's The Japanese Sword, A Comprehensive Guide, translated and with an introduction by Joe Earle (New York: Kodansha America, Inc., 1983)

I think Mr. Earle's point is even more valid for audio than it is for Japanese swords. But I have come to believe that in the audio world, the self-indulgent proliferation of words is only a symptom of something worse: what I call "joyless connoisseurship."

Oscar Wilde defined a cynic as someone who knows the price of everything but the value of nothing. I define joyless audio connoisseurship as a relentlessly reductionist drive to isolate, identify, and label acoustical phenomena, which in the process relegates the enjoyment of music to an afterthought. Joyless audio connoisseurship has as its root a vain (in both senses of that word) desire to claim to understand everything, in an effort to banish mystery from the perception of music. It is, in short, the triumph of rationalism.

I have a warning for you: The desire to have a rational explanation for everything is a profoundly emotional longing. I'll say that again. The kind of person who has a deep inner need to understand and explain everything...has a deep inner need. A nearly irrational deep inner need, in some cases. So be on your guard. Rationalists are often deeply emotionally needy people, which is why rationalism is often carried to irrationally fanatical extremes.

Without question, science proceeds (or, rather, plods along) by use of the scientific method. But radical departures in science as often as not come from the workings of the subconscious, or from observing nature. Examples are the German chemist Kekulé von Stradonitz, whose daydream of monkeys holding each other's tails provided the crucial insight into the structure of the benzene ring; and Philo Farnsworth, who conceived raster television scanning after watching a wheat field being harvested.

I do not propose that we abandon testing audio equipment in favor of daydreaming about monkeys. Test results indicating, for instance, that a certain single-ended tube amplifier amplifies less as the input voltage increases can provide a valuable insight into why some people might find its sound bewitching: such an amplifier functions as a dynamic compressor.

But such a conclusion does not mean that someone who likes that amplifier is a dupe or an unworthy person. This is because we do not listen to individual components, we listen to systems in rooms. If that amp avoids exciting resonances in a certain untreated room, that system as a whole may sound less unmusical than it would with a dynamically linear amplifier that runs smack into the room's acoustical problems.

Joyless connoisseurship seeks to cut off discussion and bludgeon dissent by invoking "scientific" propositions that may be true as far as they go, but often fail to do justice to the totality of the circumstances (footnote 1). My current favorite "scientific" party-pooper is the wet blanket some people want to throw over upconverting: "If the digital data weren't there originally, they were lost forever, so any 'improvement' you're hearing is artifactual."

Leaving aside the indisputable facts that the Compact Disc is an analog medium played back using analog means (footnote 2), my response to the Wet Blanket Clan is to paraphrase Gershwin: "It Ain't Necessarily (Digitally) So." This is because we do not listen to digital data, we listen to digital data turned into an analog voltage and then amplified to drive loudspeakers.

If some portion of the dissatisfaction many feel over conventional CD sound can fairly be laid at the doorstep of the so-called brick-wall filter used in digital-to-analog conversion, and if upconverting does move the brick-wall artifacts up one or two octaves, then indeed there is an improvement, and I would not call it artifactual. I would call it the removal of artifacts.

These ruminations were prompted by my recent investigations into the possible musical benefits of transferring irreplaceable older digital master tapes onto SACDs (footnote 3). The reductionist, rationalist snap judgment is that the original digital format's limitations are bred in the bone, case closed, keep your wallet in your pants (footnote 4).

A more nuanced approach would be to revisit the pulse-code-modulated digital recording process in a phenomenological spirit (footnote 5) and try to discern the nature of each step as it influences what is ultimately perceived.

We start by assuming a high-quality microphone feed, which means that with full-frequency program material, there will be content over 20kHz. PCM recording requires that any input signal components over half the system's sampling rate be cut off. (Unless discarded, these frequencies will be read as the difference between themselves and the sample rate, creating "alias" signals in the audioband.) So, the sampling rate itself is the first McGuffin. Obviously, a higher sampling rate is better, as it is the limiting factor in the system's frequency response. (But if frequency extension is the essential element, why have I been moved to my core by acoustical 78rpm recordings with no demonstrable content over 10kHz?)

Equally problematic are the limitations of the input (also called anti-aliasing) filter: It must pass no signal at 24kHz, yet as much as possible at 20kHz. The nearly vertical slope of such a filter (120dB cutoff over roughly the span of two adjacent musical tones) results in its often being referred to as a "brick-wall" filter. Designing such a filter is a tall order, and one that cannot be accomplished without sacrifices.

Footnote 1: The opposite of joyless connoisseurship is brainless enthusiasm, which usually manifests itself in Internet postings lauding the enthusiast's latest purchase as the best whatever he or she has ever heard. Such enthusiasts rarely let you know how many whatevers they have heard, much less the experiences that have shaped their discernment.

Footnote 2: As was conclusively shown in Stereophile seven years ago (Vol.17 No.12, pp.84-95), the pits of varying lengths in a CD's surface are analogs of digital data. The analog voltage from the player's photocell output is checked against a clock signal, and the player's logic circuit treats the result as a digital datastream. But the CD itself is an analog medium read by analog means. That is why enhancements into which I have made extensive professional investigations, such as black underprinting under CD labels and true cryogenic processing, do result in audible differences. These modifications influence the analog process of CD data retrieval as surely as vacuum hold-down systems and bedrock-coupled stands influence LP playback. Most of the top mastering engineers have had experiences in which artist-clients far removed from the land of ones and zeros have complained about test CDs made using one transfer medium (CDR vs 1630 vs Exabyte) but were happy with another: audible differences, despite bit-to-bit verification. As is said here in Rhode Island, go figure.

Footnote 3: Readers with long memories may be tempted to e-mail me about my March 2000 "As We See It." If so, please re-read that article. My concern was primarily the economics of SACD as a mass-market medium in an oversaturated and eroding marketplace for recorded music of all kinds, not just classical. And if I was so wrong, why does Tower Records need to hold press conferences announcing that they have no present intention to file bankruptcy? I had been offended by ivory-tower scribes who cavalierly said that independent labels should just eat SACD's higher costs in the interest of not stocking parallel inventories. I have always said that DSD sounds good. In view of DVD-Audio's slow-motion death wish (what was so bad about two-channel, 24-bit/96kHz linear PCM on DVD-Video, anyway?), SACD is the only generally available high-resolution music vehicle. SACD also seems poised to reach critical (but still minority-status) mass.

Footnote 4: Obviously, new all-DSD recordings, and SACD remasterings from high-quality analog master tapes, are totally different cases.

Footnote 5: Phenomenology is the brave attempt to fashion a philosophical system totally without preconceptions, while focusing on the interactions between people and perceptions. Phenomenon is the Greek word for appearance.