B ut of what essence are these materials made? This essence is the intelligence of man solidified in a way: intelligence that seeks, questions, infers, reveals, and foresees at all levels. Music and the arts in general seem necessarily to be a solidification—a ‘materialization’—of intelligence. Naturally, this intelligence, though humanly universal, is diversified by the individual: by talent that distances one individual from the others. Talent is therefore a kind of qualification—a gradation of the vigor and richness of intelligence. For intelligence is, fundamentally, the result—the expression of billions of exchanges, of reactions, of transformations of energy in the cells of the brain and the body. One could, taking astrophysics as a visual aid, say that intelligence is the form taken by the minimal acts of cells in their condensations and their movement—as happens with the particles of the suns, planets, galaxies, and clusters of galaxies born of or turned back into cold interstellar dust... To think up music as composer, craftsman, and creator, it is first necessary to study solfège, notation, music theory, and even an instrument over a long time. And since, in addition, musical creation is considered superfluous, very few people are able to attain it. Thus the individual and the society are deprived of the formidable power of free imagination that musical composition offers them. We are able to tear down this ‘iron curtain’, thanks to the technology of computers and their peripherals. The system that has succeeded at this ‘tour de force’ is the UPIC (Unité Polyagogique Informatique du CEMAMu)...”T he performance by International Contemporary Ensemble and Steve Schick of five major works—‘Psappha’, ‘Echange’, ‘Akanthos’, ‘Palimpseste’, and Xenakis’s final composition, ‘O-Mega’—is an opportunity not to be missed. It’s at Chicago’s Museum of Contemporary Art, tonight at 19:30. Anticipating this ICE performance got me to listening to the recordings I have of Xenakis’s music. I have never been lucky enough to hear a ‘live’ performance like this...
— Iannis Xenakis in a typically ‘expansive’ moment in the 1970s, quoted in ‘The Art of Music: Tradition and Change’ by William Christ and Richard DeLone.
Source-code:
Output:
X enakis’s ‘Morsima/Amorsima’ score says that ‘moros’ = ‘fate’ and ‘amorsima’ = ‘that which does not come from fate’. It was composed on an IBM 7090 in Paris, in increments/revisions between 1956 and 1962, and it was, like Xenakis’s ‘Achorripsis’, supposedly entirely stochastic (random, computer-generated). To me, this beggars belief.
F ew have ever performed a deep-dive source-code review of what Xenakis programmed, nor compared what the 7090 generated to what Xenakis actually wrote down as a published score. Oh, yes, there are people who have ported the FORTRAN code in Xenakis’s book to FTN95, and, yes, the code does run—it does produce things. It has calls to the built-in FORTRAN random-number function; it modifies global variables and accesses the system clock to accomplish a modest degree of runtime indirection and aleatorics, in the limited ways that the FORTRAN language was (and is) able to support.
B ut these source-codes do not produce things that have any close resemblance to any composition that Xenakis actually published. And the in-line commenting of the source code is terrible. The rats' nests of GOTO statements upon GOTO statements are nearly impenetrable-incomprehensible, even by the not-so-great software engineering standards of his day. A static analyzer would be the thing to use with this wild-hare code, not manual ‘desk analysis’! PRQA used to sell an analyzer called QAfortran™. I don’t know whether this is still available. I don’t see it on their website...
W e have lots of software things today that serve as composition-aids... PWGL, Vielklang, Max/MSP, Jitter, harmonization plug-ins in Finale® and other notation applications… all sorts of things. And in the 2000s [almost] nobody goes around touting their use of composition software in the way that Xenakis did. By the same token, [almost] nobody is embarrassed to say that they do use them, or to forthrightly confess the methods by which the tools contribute to the compositional process.
X enakis’s code supposedly causes the [specifications for the scored—] sounds to be generated as a directed-graph (digraph). First the [random] sequence of pitches is generated, with each note’s time of specified onset—it’s time coordinate. Then its timbre is generated (the instrument that is to play it; and its articulation: arco, pizzicato, glissando, and so on). Then the velocity of attack. Then the gradients in loudness (dynamic changes in loudness during the time interval the note is sounded... legato, accent, marcato ^, staccato, slur, portato/martelé ... sforzando, subito whatever) and in pitch (vibrato or not, glissando beginning and ending, shakes’ frequency and pitch-width). Then the duration. Not sure whether the program had anything in it to specify the (random) mode of releasing each note.
[50-sec clip, austraLYSIS, Iannis Xenakis, ‘Morsima/Amorsima’, 1.6MB MP3]
[50-sec clip, austraLYSIS, Iannis Xenakis, ‘Morsima/Amorsima’, 1.6MB MP3]
T he 7090 was a room-sized $3M, 36-bit single-CPU machine, with 32K words of magnetic-core RAM (144 KB) and a clock cycle of 2.18 μs (459 KHz). It computed at about 0.2 MIPs, 0.03 MFLOPs. [The little laptop on which I am now writing this blog, is 64-bit, 8 GB, over-clocked 3.06 GHz dual-CPU, 46.7 MIPS (Dhrystone ALU), 35.2 MFLOPS (Whetstone iSSE3).]
B ut the probabilities in ‘Morsima-Amorsima’ fate-vs-nonfate are highly unlikely. And this piece and others are too, too beautiful to be ‘lightly edited’ output from any of Xenakis’s FORTRAN source-codes that are available. Instead, there are the hallmarks of the human ear, the human hand, the Xenakisian mind. I am betting that this ‘Morsima/Amorsima’ is maybe 80% Xenakis traditional compositional method-produced and only 20% Xenakis algorithms on IBM 7090-produced—20% Fate and 80% Non-Fate, by one way of thinking about random events and chance. The score only costs €50. We should take a detailed statistical look and see!
I am not saying that Iannis Xenakis deceived anyone, exactly, regarding how his compositions were produced. He was a brilliant, flamboyant character and a wonderful and entertaining speaker, and I experienced these qualities of his, first-hand. I attended a lecture series he gave in 1972… I had been taking Computer Science courses at University of Minnesota at the time. The exuberance and expansiveness he exhibited when speaking were fabulous.
B ut I think that not enough detailed consideration has been given to the stories he told about his compositional methods—comparing his writings and interviews and software source-codes against his finished scores, to see how well (or not) Xenakis’s remarks about his methods are supported by the evidence.
X enakis’s famous book ‘Formalized Music’ and the FORTRAN source-code fragments that have been published obfuscate more than they reveal. Was he unsatisfied with the musicality of the results of his algorithms? Did he feel that he needed to extensively rewrite the computer-produced compositions to achieve a more aesthetically appealing, defensible work? What motives would lead him to withhold so many of the details? Did he have a sense of obligation to IBM Paris, who had provided resources and upon whom he depended? Was it pride? The François Bernard Mâche and the James Harley biographies are well worth studying! But none of the biographers to-date have been software engineers capable of performing detailed code-reviews, to ascertain whether or not Xenakis actually did things in the ways he claimed that he did.
I f you have views about these things—about the roles of computational composition and interactions between algorithms and human composers; and about Xenakis’s evasiveness about his actual methods—please add a comment below or send me an email. If you knew Xenakis personally and think my speculation is a bunch of hooey, please write and tell me so. If you yourself delegate to Vielklang or PWGL or other software tools part of your composerly work, I’d be delighted to hear from you, too, to learn how you use these tools to generate ideas and edit them.
O f course, if you’re able to attend the Steven Schick—ICE performance tonight in Chicago, please put a comment here, too! Thanks!
- Xenakis score pages
- Xenakis 'Morsima/Amorsima' score at Boosey
- Xenakis FORTRAN code (from 'Formalized Music') ported to FTN95.
- Bailes F, Dean R. When is noise speech? A survey in sonic ambiguity. Computer Music Journal 2009; 33:57-67.
- Harley J. Xenakis: His Life in Music. Routledge, 2004.
- McKechnie S, Stevens C. Knowledge unspoken: contemporary dance and the cycle of practice-led research, basic and applied research, and research-led practice. in Practice-led Research, Research-led Practice in the Creative Arts, Smith H, Dean R, Eds. Pp. 84-103 , 2009.
- Mâche F. Portrait(s) de Iannis Xenakis. Bibliotheque nationale de France, 2001.
- Schiege G. Iannis Xenakis: The electroacoustic work. Computer Music Journal 2007; 31:82-92.
- Smith H, Dean R, eds. Practice-Led Research, Research-Led Practice in the Creative Arts. Edinburgh Univ, 2009.
- Solomos M, ed. Présences de Iannis Xenakis. CDMC, 2001.
- Temperley D. Music and Probability. MIT, 2007.
- AustraLYSIS. Windows in Time. (Tall Poppies, 1995.)
- Varga Β. Conversations with Iannis Xenakis. Faber and Faber, 1996.
- Xenakis I. Formalized Music. Pendragon, 1971.
- Xenakis I. Science and music: An interview. UNESCO Courier 1986; 39:4-7.
- IBM Archives - 7090
- IBM 7090 page at Frobenius
- MusicFromMathematics
- austraLYSIS website
- Roger Dean page at Jazz.com
- Roger Dean page on MySpace [RecordedShapeMSPmore sensory; Tuning the Tempers Pt 2; other interesting pieces]
- Roger Dean page at Univ Western Sydney
- Dean R, Bailes F. Human understanding in imagining and organising sound: some implications of John Locke’s Essay for ecological, cognitive and embodied approaches to composition. Organized Sound, 2007;12:89-95. [200KB pdf]
No comments:
Post a Comment