March 12, 2007
Is It Live ... or Yamaha? Channeling Glenn Gould
By EDWARD ROTHSTEIN
Was that relief I felt as the piano was playing? A feeling that some worry had been alleviated or a fear quieted? Why then was it also mixed with disappointment, as if some deep yearning had been thwarted? Not yet, not yet, not yet: relief and frustration intertwined.
For months I had deliberately avoided listening. A technologically oriented, musically sophisticated company, Zenph Studios (zenph.com), claimed that it could bring the voices of the musical dead back to life. It could achieve, that is, what technology has long dreamt of: It would make light of the material world and all its restrictions.
Zenph claimed it could take a 50-year-old mono recording and distill from its hiss-laden, squished sound all of the musical information that originally went into it. It wouldn’t “process” the recording to get rid of noise; it wouldn’t pretend to turn mono into stereo; it wouldn’t try to correct things that were sonically “wrong.” Instead the claim was that it would, using its proprietary software, learn from recorded sound precisely how an instrument — a piano, for starters — was played, with what force a key was struck, how far down the sustain pedal was pressed, when each finger moved, how each note was weighted in a complex chord and what sort of timbre was actually produced.
Then it would effectively recreate the instrument. A digital file encoded with this information would be read by Yamaha’s advanced Disklavier Pro — a computerized player piano — and transformed into music. A recorded piano becomes a played piano. This would be sonic teleportation, monochromatic forms reincarnated as three-dimensional sound — not colorization but re-creation.
Zenph also announced it had accomplished this feat of technological legerdemain with one of the most remarkable recordings of the last century: Glenn Gould’s 1955 mono rendition of Bach’s “Goldberg” Variations. Gould, who retreated from performance into the private realm of the recording studio where he could splice and fiddle with sound and phrase, would be posthumously pulled back into the realm of public performance.
Gould believed technology liberated performer and listener. Here this pianist, who died in 1982, would be freed from the ultimate constraint.
And indeed, last September in Toronto, Zenph gave a public “reperformance” of Gould’s “Goldbergs” on a specially prepared Yamaha Disklavier. Zenph’s “Goldbergs” inspired a standing ovation from the audience members, many of whom knew Gould and some of whom had heard him play live. The press reports glowed.
Then one day last week Zenph — which took its name from “senf,” the German word for mustard — brought a press demonstration of its “Goldbergs” to Yamaha’s New York piano studios, playing portions of the work both on the Disklavier and from its recording, due to be released at the end of May on Sony BMG Masterworks.
Before the demonstration I returned to the 1955 recording, which I had not heard for several years. I was swept away again. This is not spiritual playing, plumbing the profundity of Bach’s meditations; it is ecstatic, uncanny in its intoxication. The recording is skittish, illuminating, thrilling and extraordinarily physical: the playing seeps into muscles as well as ears; every phrase exerts the pressure and play of dance.
John Q. Walker, Zenph’s president, knows this as well. He is a brilliant software engineer (who did important work in computer networking) and a musician who speaks of his enterprise with impassioned fervor. Last week, when he started the Yamaha instrument playing his encodings of Gould, something thrilling really did take place. The piano produced sounds that were indisputably human and unmistakably Gouldian. The playing could not have come from any other pianist.
But wait. ... Gould’s recorded piano sound is dry, as if each note were squeezed free of moisture. The phrases quiver; connections between notes are tensile, as if they were being held together by sinews. But at the demonstration the sound was often plump, rotund, even bell-like. That is partly the character of Yamaha pianos. And isn’t that a problem? Any great pianist will adjust a performance to the instrument, treating one with a “wet sound” differently from one with more sharply etched qualities, phrasing differently, even adjusting tempo. This difference in instruments limits Zenph’s claims; it also seemed to slacken the music’s sinews.
Presumably though the recording — done on another Yamaha that the piano technician, Marc Wienert, voiced to resemble Gould’s old Steinway — would have a better effect. Yet it leaves a similar impression. Is this some psychoacoustic phenomenon then, some disorientation caused by close familiarity with the old mono sound? When recordings were first becoming widely available at the turn of the 20th century, there were demonstrations in concert halls in which singers would begin a song, and a hidden gramophone with its amplifying horn would complete it. One London newspaper reported: “The most sensitive ear could not detect the slightest difference between the tone of the singer and the tone of the mechanical device.”
Bizarre. But am I experiencing something in reverse, treating sonic antiquity with reverence and not recognizing musical similarities? We all learn languages of listening, ways of interpreting reproductions, imagining full-size orchestras emerging from clock radios, ignoring hisses or distortions, compensating for flaws.
Does the new instrumentation seem less convincing because it disrupts the old familiar language of listening? I don’t think so. In Zenph’s recording, the music’s tensile line really is loosened. I admire what I hear and might not even realize what was missing without comparing, but I am not intoxicated with Gould’s exuberance or infected with his ecstatic amazement. The music is the same, yet not the same.
Of course one might say, “How could it be otherwise?” Think of the kinds of processing and analysis that had to be done: filtering out Gould’s hums or groans, isolating the sound of the piano with all its intricate overtones, taking into account the way sound was compressed or altered by every microphone, processor or wire it passed through.
Then there’s another step: “reverse engineering” the sound, as if reconstructing the instrument that created it. Then another: producing the music from yet another instrument, Yamaha’s Disklavier. And another: recording the music yet again.
The process is mind-bogglingly complex. And at every moment there are also human decisions — adjustments of the piano, musical alterations. Perhaps over time both human practice and technological possibilities will evolve further, leaving fewer distinctions. A recording by Art Tatum is due next from Zenph, along with other recordings from Sony BMG Masterworks’ rich archives.
But why all this effort? (Five man-months for a “reperformance,” as Mr. Walker explained.) Partly perhaps because contemporary sound is considered preferable and marketable. Partly because, as Zenph’s Web site points out, the great recordings of the past are passing into the public domain. The European Union allows just 50 years of protection — and this is a way of maintaining proprietary control.
But is the result really musically superior? It could only be that if there were absolutely nothing lost and every difference were an improvement; neither is the case. This is a disappointment then, though one that is exhilarating in its enterprise and promise.
The disappointment is also a relief. For had Zenph succeeded, there would have been a severe price. Had that really been Gould’s sound coming from the piano, it would have dealt a severe blow indeed to an ancient prejudice: that music, in all its complexity, is beyond the reach of the merely technical, and that it belongs, in creation and interpretation, to humanity’s ever-shrinking domain. Relief: no Gouldian robotics. Yet.
Connections, a critic’s perspective on arts and ideas, appears every other Monday.