Press "Enter" to skip to content

Not So New: Machines That Play

In early 1746, the parfumier and glovemaker Pierre Dumoulin left his native Lyon and headed north towards Germany. He was on his way to foreign lands to exhibit three remarkable objects constructed by the celebrated maker of automata, Jacques de Vaucanson. This unlikely trio was made up of two robotic musicians and a mechanical duck.

One musical figure, dressed like a dancing shepherd, could play “twenty Tunes, Minuets, Rigaudons, and Country-dances” on a pipe held to its mouth with one hand while beating on a tabor with the other. The duck appeared capable of all the movements of a living animal, the most remarkable of which were internal: after dabbling greedily at handfuls of corn offered it, Vaucanson wrote that, “The Matter digested in the Stomach is conducted by Pipes, quite to the Anus, where there is a Sphincter that lets it out.” The grains had been miraculously transformed by the automaton’s ingenious gastrointestinal apparatus into stinking excrement. The canard was later revealed to be an elaborate fraud, one so effective that it had fooled many an eighteenth-century scientist.

“Vaucanson’s Automatic Duck,” a fanciful reconstruction on paper of the innards of the Defecating Duck, Scientific American, Jan. 21, 1899 (Linda Hall Library)

Though he perpetrated this infamous anatine hoax, Vaucanson was not an across-the-board quack. After Frederick the Great was unable to convince him to come to Berlin and become his personal mechanic of marvels, Vaucanson went to work for King Louis XV of France, and he invented a mechanical loom, which proved to be a crucial technological impetus for the Industrial Revolution.

Vaucanson’s first and most famous musical invention, the one he triumphantly presented to the Royal Academy of Sciences in Paris in 1738, was a faun that played the flute. Nearly six-and-a-half feet tall, the construct was not a stiff, unmoving machine, but astonishingly realistic: it had fingers with leather pads that stopped and unstopped the holes of the flute it held in its hands; it had lips and a tongue and a throat through which came a variable breath. That Vaucanson was able to create a convincing facsimile of human performance on this most difficult and nuanced instrument, one that required the life-giving, God-given gift of breath, was a testament to the masterful engineering concealed within: three sets of bellows produced different wind pressures; a series of levers and pulleys allowed the lips to protrude and to change the size of the windway, controlled the action of the tongue, and the movement of the fingers—all these carefully constructed parts, engineered to minute tolerances, were governed by a precisely pinned cylinder that delivered a diverse repertory. Vaucanson claimed that his figure could enact motions—and music—comparable to “those of a Living Person.”

The mechanical flautist had startling implications for human musicians. Writing in 1752, Johann Joachim Quantz, flute master to Frederick the Great, claimed rather defensively that, however impressive it might be, the faun’s impeccable technique only highlighted what it lacked—the ability to perform a piece with the “proper fire.” Whereas human musical utterances were, at their best, stoked with emotion and ignited by imagination, machines were cold and unfeeling.

One of Johann Sebastian Bach’s ardent supporters, the theologian Johann Michael Schmidt, was equally dismissive of Vaucanson’s invention, though the critique of it he published in 1754 also revealed his fearful unease about technological encroachment:

Not many years ago it was reported from France that a man had made a statue that could play various pieces on the flute, placed the flute to his lips and took it down again, rolled its eyes, etc. But no one has yet invented an image that thinks, or wills, or composes, or even does anything at all similar. Let anyone who wishes to be convinced look carefully at the last fugal work (The Art of Fugue) of Bach … I am sure that he will soon need his soul if he wishes to observe all the beauties contained therein, let alone wishes to play it to himself or to form a judgment of the author.

What would Quantz and Schmidt have said about AI, and with it modern recording and sound-manipulating and generating technology—about all those precise, nuanced, and infinitely repeatable and transformable “performances” accrued from the exponentially expanding stores of digital detritus? Part of Quantz’s objection was to the unvarying approach of the faun’s “interpretation”—it always played a given piece the same way, since its actions were determined by the pinning of the cylinder. But what Quantz never answered is why the person responsible for pinning the cylinder could not produce a single performance that follows principles of human good taste, elegance, and even fire.

The AI successors of Vaucanson’s flute player are now calling the tune.

It is now not hard to imagine digital performers and performances that “think” for themselves, ones that can artfully navigate their way through the massive quantity of information stored within, ones that are capable of changing nuances of tempo, dynamics, and ornament in infinite and pleasing combinations, perhaps calibrated to the algorithmized preferences of the human audience whose data they crunch as they hurtle through a uniquely riveting cadenza. Our artificially intelligent performer could even learn to throw in an occasional mistake—the artfully placed error, the human blemish that brings the beauty of the interpretation into even greater relief. Even in the good old analog days, some recording editors already did that, purposefully seeking out one, or even two, elegant miscues to work into the final, collated take.

Does all this portend, at last, the devoutly-to-be-wished-for destruction of the virtuoso ego? One way to accomplish this might be through a grand digital synthesis, an artificially intelligent alloy of Schnabel, Rubinstein, Horowitz, Schiff, Lang Lang, and Yuja Wang. Might a new Vaucanson devise a musical automaton with the left hand of Pollini and the right hand of Richter? Or perhaps a digital test-tube baby, the techno-love child of Martha Argerich and Van Cliburn, his data exhumed and made now to rest not so peacefully.

I’ll bet that my master-mix version of Beethoven’s Hammerklavier Sonata gathered from the most famous twenty pianists active today could fool any international competition jury and might even be better than the sum of its parts, thus defeating the Romantic notion of unique, individual genius. Deep Blue V beat Kasparov at chess long ago; Schroeder XXVI could well clear the field of human competition and then give a charming AI-powered post-concert interview lighted by the incandescent humor of a Schnabel.

The God-loving philosophers and musicians of yore could claim that one can never edit in, or artificially generate, those ineffable elements that make up a convincing performance. This is a version of the Quantz and Schmidt critique of Vaucanson’s flute player: that the soul will shine through.

Yet even old-fashioned recordings have long been battering away at this metaphysical Maginot Line. More than any other musician, Glenn Gould was the pioneer of obsessive editing in the hermetic confines of the studio, all in pursuit of the perfect and, once achieved, immutable interpretation. He brought this approach to bear most famously—or infamously—on Bach’s keyboard works. Gould was the pre-digital prophet who first worked the dubious miracle that has now become standard operating procedure. Faultlessness became the ideal, one more easily attained through digital means. As has often been pointed out over the last decade, this standard has a controlling effect on live performances, which inevitably strive to match these levels of exactitude at considerable cost to the kind of variety and warmth Quantz and others cherished. The circle of live performance and digital recording spiraled downward towards dubious perfection.

Then there is the Canadian piano virtuoso Marc-André Hamelin, whose digital technique (I mean his fingers, not his post-production editing chops) and his capacity for synthesizing various keyboard styles are so colossal that artificial intelligence and robotics might not overtake him for another few years—or maybe just months. This much is obvious from his jaw-dropping, brain-frying, finger-busting Twelve Études in the Minor Mode. The set is an homage in title and concept to that of the nineteenth-century French piano and organ genius Charles-Valentin Alkan, whose fiendishly difficult music Hamelin also tosses off with dumbfounding brilliance. Hamelin also happens to be the name of the German town from which came the fairy tale about the rat-catching Pied Piper, though the Canadian turbo-pianist could never be satisfied with a single, fluting line. With this Hamelin it’s an ebony-and-ivory blitzkrieg.

Finished in 1992 when Hamelin was in his early thirties, the first of his twelve test pieces is a “Triple Étude,” riffing simultaneously on no less than three different Chopin piano works, each supremely difficult in its own right. Hamelin’s two-minute sprint starts off innocently enough, but by the halfway point it’s as if the pianist has sprouted two more arms, each equipped with five super-accurate fingers.

The nonchalance is terrifying, and if he were a machine (I’m assuming he isn’t), one might shudder, even weep as the frolic fizzes on.

The miraculous feat ends in an expansive, pianissimo chord in A major, the minor escapade flashing a mischievous grin as it evaporates into thin air, the whole thing so unbelievable it must have been a mirage. From here, the collection gets even more superhuman.

Hamelin’s digits move at bullet-train speed, but do they move the spirit? The music is fun but freakish, the man is many phenomenal pianists in one.

Will Hamelin retain his superiority of intellect and dexterity over robotic, machine-learning competition, or has he already lost it? The centuries-long virtuosic drive for more and faster has met its match—indeed, is responsible for making its supposed adversary.

The mania turns back on itself and, in doing so, shows the way out.

What musical culture needs—what the world needs!—if it is to fend off the machines and maniacs is the real shit, not Vaucansonian tricks: honest imperfection, unlikely imagination, and more humanity.

(David Yearsley is a long-time contributor to CounterPunch and the Anderson Valley Advertiser. His latest albums, “In the Cabinet of Wonders” and “Handel’s Organ Banquet” are now available from False Azure Records.)

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

-