Victor Shepardson

Full Text

In this typically abstract, discursive essay George Lewis discusses notions of improvisation around his machine improvisor Voyager.

...improvised music privileges the role of musical sound as a carrier for history and cultural identity. Improvised music deemphasizes Western “classical” notions of form and structure in favor of exchange of cultural and social narratives. Informed listening to improvised music, including my own work with computers, will involve attention to this process.

The notion of virtuality explored by Voyager... contradicts the idea that computer-driven musical activity is primarily about the technical replacement of ‘real’ musicians with their virtual counterparts. Instead of a virtuality that attempts to hegemonize the physical, the goal is one where virtuality and physicality interact to produce a hybrid that strengthens on a human scale.

Voyager is a computer improvisor which analyzes acoustic signal from other agents and produces MIDI to control an orchestra of synthesized instruments. It’s placed on equal footing with a human improvisor; all communication with the machine is through sound. Lewis’s use of traditional instrumental timbres with Voyager emphasizes the machine’s musical personality at the note level (rather than camouflage it with alien sonorities). Voyager has also been hand-coded by Lewis himself, deeply embedding his own assumptions and intuitions in its behavior. As he concludes:

Full Text

Video

This NIME paper by Eric Sheffield and Michael Gurevich presents a mechanically actuated multi-percussion instrument. The authors describe a pair of found-object percussion instruments which have been electromechanically coupled. They are: a ceramic tile activated by a robotic brush; and an aluminum bar struck by a trio of solenoids. The sound of each object is transduced by a magnetic pickup and used to control actuation of the other. Each object is also supplied a human-operable gate which enables output from the pickup. The resulting system is thus primarily under control of a human performer, with standard percussion gestures as input.

Full text

This essay by Christoph Cox relates the cosmology of Leibniz to the concepts of sound, signal and noise as a way of understanding sound art.

Cox begins by rejecting two conceptions of signal and noise: first, that signal is primary and noise is some secondary corruption of signal. Second, that the status of signal and noise are relative, with signal being that which is intended to have meaning. Cox instead investigates noise as something cosmic, fundamental, inevitable.

Leibniz imagines a universe wherein every element is mutually linked in a vast web of causation. Therefore, the content and behavior of matter at each point in space and time reflects the order of the entire universe. 

Similar ideas have been developed by David Bohm, who calls this the implicate order.

For Leibniz, as a consequence of this idea the entire vastness of the universe is available, through the senses, to the human unconscious. 

Cox clarifying Leibniz: 

When I walk along the seashore, my perception of ‘the great noise of the sea’ is clear; that is, it is fully and powerfully audible. But it is also confused, since I hear this sound as a mass and don’t distinguish its elements – the individual waves – which remain obscure. Yet I must in some sense hear the individual waves, otherwise I could not hear the aggregate. Hence the sound of each individual wave must be distinct for me, though in an unconscious and, hence, obscure sense. What is clear, then, is also confused, and what is distinct is also obscure

I don’t think that’s a given physically or physiologically speaking. Why should we assume that the sound incident on one’s eardrum contains complete information about the crashing waves? (A deterministic universe need not be time reversible.) And why suppose it would escape destruction in the mechanism of the ear? The human ear is not a perfect instrument; it’s part of an informational funnel selecting only those acoustic features most relevant to its own survival and reproduction. (It’s worth noting that Leibniz died almost a century before Darwin was born.)

The idea that the whole of the implicate order is available to the unconscious (important enough to Cox to put in the title of the essay) seems tied up with Leibniz’s theology. All is signal to God; noise the signal too enfolded for humans to understand. At this point Cox takes a different view. I am probably missing something here because I can’t really follow this passage to its conclusion:

noise is the set of sonic forces that are capable of entering into differential relations with one another in such a way that they surpass the threshold of audibility and become signal. Noise and signal, then, are not differences in degree or number but differences in kind, distinct domains. Noise is no longer merely one sound among many, a sound that we do not want to hear or cannot hear. Rather, it is the ceaseless and intense flow of sonic matter that is actualised in, but not exhausted by, speech, music and significant sound of all sorts.

I think where we end up (though I wish Cox would state it more clearly) is that the signal/noise relationship depends on the attention of the listener. All is noise; signal is that which humans attend to and understand.

full text

This paper describes an event exhibiting recent work in the field of Musical Metacreation and proposes a taxonomy of metacreative systems.

This classification depends heavily on the notion of a musical gesture, being “any musical idea that has a beginning and an end”. This definition mostly passes the buck from “gesture” to “idea” which if anything is more vague. One might define gesture recursively as a superposition of gestures and sound objects. I’m using sound object to mean an acoustic event which streams together and seems to a human observer to represent a single source. Then, a gesture is built up from particular atoms in a particular temporal configuration; this allows for gestures which are single sounds (e.g. a snare drum hit), gestures which are patterns of sounds (e.g. a drum beat), and gestures which are patterns of other gestures (e.g. a sequence of increasingly loud but otherwise unspecific snare drum textures)

Anyway, the authors present a taxonomy which is hierarchical, from less metacreative to more:

1. Independence

2. Compositionality

3. Generativity

4. Proactivity

5. Adaptability

6. Versatility

7. Volition

Levels 1 and 2 involve processing of human gestures in a not completely predictable way. An independent system can do this to single gestures, while a compositional system can relate multiple gestures.

Levels 3 and 4 deal with the production of novel gestures. A generative system produces novel material when triggered by a performer; a proactive system can choose when to trigger itself. That is, it reacts in a creative way, responding to but not merely echoing what it hears.

Level 5 is more sophisticated that a generative system or has more agency than an proactive system; an adaptive system is capable of substantial evolution in its behavior over time due to internal or external conditions. Thus it can produce interesting results without human intervention. This level is divided into parts a) and b), which distinguish between holistic systems with only internal state, and agent based systems wherein each agent faces the external state given by the other agents.

Level 6, a versatile system, generalizes level 5 to be “without predefined stylistic limits”. A versatile system is capable of determining large-scale musical form; it could be said to be operating as an artist.

Level 7, a system with volition (a volatile system?) decides for itself the conditions to compose in; it operates as a person with needs and desires outside of its role as a composer.

The authors do not claim the existence of any system above level 5. I think their hierarchy is mostly sound, though levels 3 and 4 seem more parallel like 5a and 5b. For example, a generative system might reach level 5 without the power of choice implied by level 4, as in a natural process like rainstorm. Even a totally procedural work (by which I mean one which is determined at each instant entirely by a set of parameters and the current time) may appear to evolve substantially due to its rich mathematical structure. There is clearly a spectrum between each of these levels since 1 and 2 are based on the fuzzy notion of gesture and 6 is based on that of “stylistic limits”. (Are many human artists really without stylistic limits?)

This is one of three prefaces in my copy of Formalized Music; it's the original preface to the French Musique Formelles, which is the first half of Formalized Music. I think Xenakis gets at something very important here:

"...consider sound and music as a vast potential reservoir in which a knowledge of the laws...and structured creations of thought may find a completely new medium of materialization..."

"For this purpose the qualification 'beautiful' or 'ugly' makes no sense for sound...the quantity of intelligence carried by the sounds must be the true criterion of the validity of a particular music"

At this point he's dangerously close to defining value as intelligence despite a crude understanding of what intelligence even is. But! The critical bit follows:

"This does not prevent the utilization of sounds defined as pleasant or beautiful according to the fashion of the moment, nor even their study in their own right...Efficacy is in itself a sign of intelligence"

Yes! This is such a powerful idea. Xenakis gets it. Beauty may be arbitrary; but it's totally crucial if you intend to make art for humans of your own time and place. It's also critical to separate intelligence from intellect: intelligence is the important quantity, and should fold in emotion with all other means of understanding the world.

Got nonlinear oscillator bank to work!

random envelope filter for spatialization experiments

update bendy to v1.1

Psychoacoustic Phenomena In Musical Composition and a partial interview with Eliot Handelman

The first link above is to a 1977 essay with a modern foreword; the second is part of an unpublished interview from 1991. They deal with many of the same questions, ones which Amacher investigated extensively in her work.

Amacher famously explored otoacoustic emissions, sounds produced by the ear. There are a few kinds of OAEs. Spontaneous OAEs are the sound of the ear’s active mechanisms straining to hear tiny noises. Evoked OAEs are the ear’s reaction to specific stimuli. Amacher worked with the evoked difference tones produced by the ear’s amplification system in response to very loud pure tones. These are what Amacher exploited in Synaptic Island and other works. When this music is played through loudspeakers at a high (near painful) volume, a new contrapuntal layer emerges which sounds on the ear or inside the head (it doesn’t work with headphones).

Amacher was interested in the interaction of these “interaural” sounds with external sounds in the air. Though Wikipedia pedantically notes that OAEs are physiological as opposed to psychoacoustic phenomena, she relied on those too. The rapid arpeggios of Synapse Island separate by auditory streaming in the same way as Bach’s cello suites. The result is a personal, mercurial, space filling sound sculpture you can never quite take in all at once. It’s music to remind you that you’re made of meat; music which is at once beautiful, uncomfortable, and alien.

Amacher clarifies the power of electronic and computer technology to the composer: “since I was able to work with electronically produced sound it was possible for me to make these discoveries, experientially... to enhance the so-called psychoacoustic responses in composing it is first necessary to learn to recognize them. And this can only be achieved experientially, unlike most music which can be imagined mentally without having physically experienced it”.

YouTube

Concret PH was composed by Iannis Xenakis in 1958 for the Philips Pavilion at the Brussels World Fair. The piece is a tape collage of a single ember popping in a fire. It’s brief, approachable, and full of brutally high frequencies. Xenakis layers the sound at different speeds and positions in space. One first hears a dense bed of embers somewhere in the distance. More sparse, higher pitched embers begin to stick out, sounding right on the speaker. Pitched down embers clank past, revealing their internal structure. The constitution of the ember-cloud changes subtly: it turns to a random-sounding collection of pitches which resembles the spectral noise in a digital vocoder. Noise seethes at the high edge of hearing. It’s over too soon for listener, through probably not for composer splicing tiny bits of tape together.

full text (pdf, possibly paywall)

This 2003 article by Tara Rodgers is broad defense of sampling as a field of musical expression. Rodgers identifies sampling as folk and Afrodiasporic traditions of quotation meeting the sound-transformation of musique concrète. She goes on to falsify the common assumptions that sampling represents theft in place of creativity or automation in place of performance. A sampler is like any other instrument in that its performance can be expressive and its technique deep; a sample is like any other sound in that it can be selected for its conceptual weight or aesthetic character.

Rodgers offers a few examples: Le Tigre using samples for their semantic content and evocation of a time and place; Matmos forging a conceptual unity between surgeon and patient, producer and sound. She challenges the notion that sampling represents a fetishism of technology or a poverty of ideas: glitch music explores the unintended behaviors of technology; a lo-fi approach to sampling can reject the pristine or expensive as a political statement.

She also makes the point that sampled music is ripe for social analysis. Woven into the article is a feminist critique of the particularly male-dominated house genre and the double standard for female producers working with a glitch or lo-fi aesthetic.

full text (pdf)

In this essay from 1962, Karlheinz Stockhausen talks about the unity of sounds as acoustic phenomena and exploring the space between rhythm and pitch. He begins by lamenting the decorrelation of rhythm, timbre, dynamics and pitch in western classical music. These are typically viewed as mutually orthogonal dimensions of musical space; generally, one can vary a musical idea timbrally without affecting its melody or increase tempo without affecting the pitch. Under an acoustic model of musical sound, Stockhausen points out, it’s possible to explore liminal regions where a fast rhythm becomes a pitch, or a resonance which had determined a pitch now gives a timbre.

At this point he gets sidetracked by a very awkward explanation of the acoustic basis of pitch in the time domain. I think this was motivated by the specific piece he discusses later; but it’s a bit misleading. Ol’ Karlheinz gets no points for ignoring the sine-wave decomposition of sound a century after Helmholtz. (It’s also unclear how much is lost in translation here.)

Anyway, the meat of the essay is Stockhausen’s explication of his piece Kontakte (he’s discussing the version for electronics alone, but I can’t find it on the internet). In Kontakte, he’s specifically exploring the relationship between rhythm rates and audio rates. The electronic parts of the piece are scored by a pair of time envelopes: the rate of an impulse train, and the center frequency of a resonant filter. While the pulses are at rhythm rate, the filter colors them. Make it sharp enough, and you’ll hear its center frequency as a pitch. When the pulses speed up above 20 Hz, you begin to hear them as a low pitched tone. That pitch dominates, and now the filter is again coloring the timbre. 

Stockhausen seems to find the division of music into timbre, pitch, and rhythm very arbitrary, but it’s not really; the intervening half century of research in psychology and neuroscience have shown that they’re a product of the specific character of human auditory perception. One can imagine an alternate musical tradition with an interest in these liminal spaces developing siren- and klaxon-like acoustic instruments which could make the sounds in Kontakte. (Is there such a tradition? That would both make Stockhausen’s point and somewhat undercut the novelty of this work). Exploring these liminal spaces is very interesting; I just find it more compelling when motivated by psychoacoustics rather than straining against it.

full text (pdf)

This short essay by Brian Eno appears in the book Audio Culture: Readings in Modern Music by Cox and Warner. Eno applies the concept of variety, from cybernetics, to the understanding of experimental music. Variety is defined as the range of behaviors of an organism or system. A system with very low variety is predictable; it always does the same thing. A system with very high variety might do anything; it’s impossible to predict or control. Interesting behavior occurs between these extremes: discernible pattern without total predictability. It’s no coincidence that life exists between the forces of natural selection (what Eno calls a variety-reducer) and mutation (a variety-increaser).

Eno argues that the classical orchestra and western classical music in general tend to suppress variety, except in the composer. Strict notation, performance by trained musicians, a hierarchically structured orchestra – all serve to reproduce the composer’s specific vision as closely as possible. Individual difference does show up in one other place, though: timbre. The replication of parts in a large orchestra creates its unique timbre thanks to individuals with different vibrato, timing, and so forth.

Expressing the roles of the composer and orchestra in terms of variety raises the question of what alternatives there might be. The primary example Eno offers is Cornelius Cardew’s Paragraph 7. Paragraph 7 revolves around variety in the behavior of its performers. It’s a flat hierarchy, where each musician chooses pitches to match the other performers. Eno makes the point that the Cardew piece exists on an organizational spectrum: at one end is the anarchy of free improvisation, at the other the rigidity of the classical orchestra. In the middle you have Cardew gently directing the behavior of individuals; or a rock band with one member writing the songs but the others writing their parts; or a jazz ensemble with control constantly shifting to a different soloist, and so on.

All this gets very mixed up once a computer can simulate the orchestra and we start programming computers to behave like individuals. The orchestra evolved as an instrument to express variety in the composer; it was the only way to realize such rich, visceral, space filling sounds. Today we have many more practical tools (with their own advantages and disadvantages, of course). Why write strictly notated music for humans today? Given a room full of expert musicians wouldn’t you be crazy not to delve into their individual differences? 

pdf full text

This paper by Dave Burraston and Ernest Edmonds is a survey of applications of cellular automata to sound circa ten years ago. They include a brief overview of algorithmic and generative music, and a technical introduction to cellular automata. They go on to talk about many projects mapping cellular automata to symbolic (usually MIDI) music. A few more direct applications to synthesis are also described.

Included is a useful abstraction of the practice of algorithmic music, with the artist as a mediator between some process and an immanent work of art. There is feedback from the artist to process as s/he redefines it, and the work to the artist as the work takes shape. This paper mostly stays out of the business of asking why pursue generative art. I think many artists would tell you that they are pursuing the beauty of nature in the same way as the photographer or the landscape painter. Or for that matter, the mathematician, astronomer and biologist. In a sense, generative methods of artmaking represent godlike powers of creation: entire worlds at the stroke of a key. Yet they also admit the freedom to merely observe what was already there, waiting to be noticed.

Cellular automata are formally defined by a set of possible states, a uniform connection scheme, and a transition function. A collection of cells are each connected to a local neighborhood of other cells. For example, an infinite chain of cells each with a left and right neighbor. Or, a finite square grid of cells each with a north, south, east and west neighbor. Each cell is in one of the possible states at any given time. The next state of each cell is calculated by the transition function according its current state and those of its neighbor cells. Most of the automata discussed here are variations on either Stephen Wolfram’s 1-D automata, or Conways’s ever popular game of Life.

Surprisingly few of the works described here are easy to find online. Some are installations with no video documentation; some are projects whose pages have gone offline; some are software for obsolete operating systems. Getting them up and running would be an archival task of its own. Cellular Grid Machine has a newer version which, disappointingly, costs money. The description on that page is worth reading, though.

Here’s a nice Life sonification toy I found on the internet. It’s large enough for a little complex behavior and the process is very audible.

I also found this example of Life controlling a granular synth. This one does a great job at representing chaos and gliders.