«NG SO R U U T N P D CA k k at z HOW T EC ar H m N O O L IC G Y S MU HA SC HANGED UNIVERSITY OF CALIFOR NIA PRESS BERKELEY LOS ANGELES LONDON ...»
Common sense suggests that this solo was ﬁxed long ago, on the day drummer Clyde Stubbleﬁeld recorded it as part of James Brown’s 1970 R&B song “Funky Drummer, Part 1 and 2.”1 Yet this two-second sequence enjoys a promiscuous, chameleonic existence. Accelerated, equalized to sound muﬄed and distant, and repeated continuously in Eric B. and Rakim’s “Lyrics of Fury” (1988), it takes on a menacing tone, matching the intensity of the rap. Similarly looped, but slowed slightly and placed underneath a haunting folklike melody, it occupies a completely diﬀerent sound world on Sinéad O’Connor’s “I Am Stretched on Your Grave” (1990)— that of the Irish lament. It masquerades as a reggae beat in Sublime’s “Scarlet Begonias” (1992) and turns wistful in George Michael’s pop ballad “Waiting for That Day” (1994). In each example, and in scores of others that appropriate Stubbleﬁeld, something of the original sound is maintained, yet its meaning changes in every new setting.2 The multiple incarnations of Clyde Stubbleﬁeld’s “Funky Drummer” arise from the practice of digital sampling, a form of musical borrowing in which a portion of one recording is incorporated into another. Since the 1980s, musicians of every stripe have embraced the technology. Their work raises a host of questions, from the aesthetic and the technical to the ethical and the legal.
W H AT I S D I G I TA L S A M P L I N G ?
Digital sampling is a type of computer synthesis in which sound is rendered into data, data that in turn comprise instructions for reconstructing that sound. Sampling is typically regarded as a type of musical quotation, usually of one pop song by another, but it encompasses the digital incorporation of any prerecorded sound into a new recorded work. The equipment used to create samples varies widely, from traditional-looking keyboards to purpose-built machines dominated by buttons, knobs, and sliders that look nothing like musical instruments, to software used on personal computers. Regardless of the gear, on the simplest level sampling works like a jigsaw puzzle: a sound is cut up into pieces and then put back together to form a digitized “picture” of that sound. When a sound wave is digitized, using what is called an analog-to-digital converter (ADC), it is not reproduced in its entirety; rather, select “samples” of the wave are assigned binary numbers. Each of these numbers represents the amplitude, or height, of a wave at a given point. When a sound is reconstructed, a digital-to-analog converter (DAC) emits voltages corresponding to each 138 M U S I C I N 1s A N D 0 s of these binary numbers. When all of these various voltages are emitted in a particular order, the result very closely approximates the original. This may seem to be an odd way of reproducing sound—breaking it down and then putting it back together—but in fact it works very well. At present, the standard sampling rate is 44,100 Hz, meaning that every second of sound that is sampled is cut into 44,100 slices; typically, each of these slices is given a sixteen-digit binary number, which allows for extremely ﬁne gradations (216, or 65,536) in measuring the amplitude of a wave. Sampling can therefore be fast and ﬁne enough so that the human ear perceives a continuous and faithfully rendered reproduction.
The advantage of digitization is that sound, once rendered into data, can be manipulated in a variety of ways down to the smallest details. Tempo and pitch can be increased or decreased in any increment, and the two can be manipulated independently. (In the predigital age, when the speed of a recording was increased, the pitch rose, and when the record slowed, the pitch fell. Think of the sound of a phonograph switching from 33⅓ to 45 rpm or vice versa.) Sounds can be reversed, cut, looped, and layered; reverberation can be added; certain frequencies within a sound can be boosted or deemphasized. Noise can be removed to make an old recording sound pristine, or even added to make a pristine recording sound old, as can often be heard in recent popular music.3 All of these manipulations can be visited upon any sound, musical or otherwise, and on any length of sound that can be recorded. A sample can be a fraction of a waveform, a single note from an instrument or voice, a rhythm, a melody, a harmony, or an entire work or album. Although sampling, particularly when done well, is far from a simple matter, the possibilities it oﬀers are nearly limitless.
As a form of musical borrowing, the roots of digital sampling reach back
more than a millennium. Consider just the Western musical tradition:
medieval chants freely incorporated and adapted melodic patterns from earlier chants; dozens of Renaissance masses were based on the melody of the secular song “L’homme armé ”; a similar craze raged centuries later when composers such as Berlioz, Liszt, Rachmaninoﬀ, Saint-Saëns, and Ysaÿe “sampled” the chant Dies irae (“The Day of Wrath”) in their instrumental works; Bach reworked Vivaldi’s music; more than a century later Gounod returned the favor, adding a new melody to Bach’s Prelude in C Major and calling it Ave Maria; Mahler cannibalized his own earlier vocal M U S I C I N 1s A N D 0 s works in several of his symphonies; Ives quoted George M. Cohan’s “Over There” in his song “Tom Sails Away”; Bartók parodied Shostakovich’s Leningrad Symphony in his Concerto for Orchestra; and so on and on.
Yet isn’t there something fundamentally diﬀerent between such traditional acts of borrowing and digital sampling? It is sometimes said that while a quotation is simply a representation of another piece, a sampled passage of music is that music. But that depends on what the meaning of “is” is. Consider a conventional example of musical quotation: in the third movement of Luciano Berio’s Sinfonia, an enormous ﬁve-movement work for orchestra and vocalists from 1968, the composer quotes music by Brahms, Debussy, Hindemith, Mahler, Ravel, Schoenberg, and Strauss, among many others. These quotations are notational—that is, Berio reproduces not the sounds themselves, but the instructions for recreating them.
The quotations are only complete when performed. Digital sampling also involves symbols—1s and 0s instead of the various lines, dots, and squiggles of traditional notation. As a standard textbook on computer music explains, “What computers manipulate is not sound itself but representations of sounds.”4 Therefore, if sampling represents sound, we cannot say that a sampled passage of music is that music.
But if sampling does not diﬀer from traditional musical borrowing in kind, it certainly diﬀers in degree. Consider a hypothetical quotation, in which the score of an otherwise original work notates the two-second “Funky Drummer” solo. At most, only a dozen or so instructions (in the form of various symbols) would be used: several to indicate the parts of the drum kit (bass, snare, tom-tom, hi-hat, etc.), a handful for the duration of each note, and a few for dynamics, accentuation, and meter. But the equivalent digital sample would require nearly a hundred thousand distinct instructions, a level of speciﬁcity impossible to notate. With all of these instructions, so much more can be indicated: the sound of a particular drum being hit with a certain amount of force using a speciﬁc stick, or the exact number of milliseconds a note enters before or after the beat.
Moreover, the sonic aura surrounding the sound can also be captured. By “aura” I mean two things: the reverberation that imparts a sense of space, and the slight but constant ambient noise—a patina, perhaps—that is a by-product of imperfect recording ﬁdelity. Digital sampling oﬀers the possibility of what I would call performative quotation: quotation that recreM U S I C I N 1s A N D 0 s ates all the details of timbre and timing that evoke and identify a unique sound event, whether two seconds of Clyde Stubbleﬁeld’s drumming or the slow, unsteady tapping rhythms produced as I type this sentence. In other words, traditional musical quotations typically cite works; samples cite performances. As we will see, it is the possibility of performative quotation, including the ability to manipulate those sounds, that sets sampling apart from traditional quotation and has led to some astonishingly creative works of modern music.
T H E U N C O M M O N PA R L A N C E O F PA U L L A N S K Y
I was sitting on a plane just before takeoﬀ when an announcement came over the loudspeaker. It was no doubt the usual welcome, but for some reason I could not quite understand what the attendant was saying. At ﬁrst I thought the loudspeaker was faulty, and then I put the diﬃculty to the noise of the engines. I leaned forward, closed my eyes, and concentrated, yet I still could not make sense of the words. My frustration mounted, but then suddenly I could understand her perfectly. I quickly realized why, much to my chagrin: I was on a KLM ﬂight to Amsterdam, and it dawned on me that the attendant had given the announcement twice—ﬁrst in Dutch, and then in English. Dutch, at least to my ears, sounded quite a bit like English.
But it did not occur to me that she was speaking a diﬀerent language. Rather, it seemed as if she were using all the basic and familiar sounds of English, but in a completely unfamiliar (and rather maddening) way.
I am reminded of this incident when I listen to Notjustmoreidlechatter, a 1988 work by composer Paul Lansky. Lansky, a professor of music at Princeton University, creates his music almost exclusively with computers, and the eight-minute Notjustmoreidlechatter is one in a series of works in which he digitally manipulates speech—English speech—to create fantastic musical textures in which semantic meaning is tantalizingly out of reach.5 Lansky has long been interested in using the computer to transform the everyday into music, or perhaps to extract the music from the everyday. He ﬁnds inspiration in unexpected places—conversations, highway traﬃc, a bustling shopping mall, his own kitchen. Rather than sampling preexisting works (as the other composers discussed in this chapter do), he mines raw sonic material; moreover, these works bear little conM U S I C I N 1s A N D 0 s nection to the world of traditional performance. Lansky is thus presented with a distinct compositional challenge: How does a composer write music that lives only on recordings? That is, how does one write a work that not only must stand up to exact and frequent repetition, but must also create its own self-suﬃcient world outside the familiar traditional concert venues? Lansky answers these challenges in the form of Notjustmoreidlechatter.
Notjustmoreidlechatter (Track 10 on the accompanying CD) opens with what one might take for the Babel of legend. Countless unintelligible voices—high, low, fast, slow—bombard the listener from every direction.
Heard on headphones (perhaps the “natural” venue for such a piece), the voices seem to be inside one’s head, bouncing and darting chaotically. In fact, we are hearing only one voice, that of Lansky’s wife, Hannah MacKay.
MacKay is reading from chapter 25 of Charlotte Brontë’s Jane Eyre, in which Jane tells Rochester of her unusual dreams. The subject seems appropriate to the piece, for the disembodied voices have an unreal, otherworldly sound. While MacKay’s voice is digitally multiplied, fractured, and transformed so that no single word is long or distinct enough to be understood, it is still possible to pick out recognizable syllables or phonemes. Here Lansky strikes a balance between familiarity and strangeness, in which listeners instinctively “squint” their ears, as Lansky puts it, in an attempt to understand what is being said. (Much as I did when I tried to make sense of Dutch on the KLM ﬂight.) This is a canny compositional strategy, for it not only encourages attentive listening but also addresses the problem of repeatability. Even the most careful scrutiny will not reveal the text, but with every successive hearing the listener cannot help trying to extrapolate meaning from these verbal scraps. Here Lansky exploits the human tendency to ﬁll in missing or unclear information to form whole structures. This is the same tendency that leads listeners to misinterpret indistinct song lyrics, even if the result makes little sense, for nonsense seems always to be more tolerable than uncertainty. (Examples of misheard lyrics are legion: “Excuse me while I kiss this guy,” instead of “Excuse me while I kiss the sky”; “The ants are my friends, they’re blowin’ in the wind,” instead of “The answer, my friends...”; and so on.) Play Notjustmoreidlechatter to a group of listeners and you will ﬁnd that they all think (and even insist) that they hear particular words, though few if any will agree on what is being said.
142 M U S I C I N 1s A N D 0 s Lansky responded to the repeatability issue in another way as well. Using what he describes as stochastic mixing techniques, he essentially instructed a computer to determine certain aspects of the chattering at random. As Lansky has explained, the purpose of this unpredictability is to compensate for the ﬁxity of the recorded medium, and in doing so simulate the
spontaneity, the “danger,” of live performance: