Page  00000001 VARESE'S POEME ELECTRONIQUE REGAINED: EVIDENCE FROM THE VEP PROJECT Richard Dobson, John ffitch, Kees Tazelaar Department of Computer Science, University of Bath ABSTRACT The Poeme Electronique in the Philips Pavilion at the World Fair of Brussels 1958 was a unique experience, the very first multimedia project to involve a sense of the total experience of vision and sound. Realized by three relevant personalities of contemporary culture, Le Corbusier (conception after Kalff, coordination and visual show), Varese (organized spatial sound delivered over 350 loudspeakers), Xenakis (architecture of the pavilion), the Poeme Electronique has never been repeated after the Philips Pavilion has been pulled down after the Expo. The Virtual Electronic Poem (VEP) project is reconstructing the lost artwork with virtual reality techniques. In this paper we consider the technical aspects of the reconstruction of the audio part of the Poeme by describing the investigations made on the piece and by connecting the Poeme to the whole Varesian production. 1. INTRODUCTION The VEP project' is aimed at reconstructing in Virtual Reality the multimedia show of the Poeme Electronique as performed inside the Philips Pavilion in 1958. In order to do this we need to construct the complete audio material before it can be treated to the spatialisation implicit in the environment. In our investigation we have been strick by how all the available recordings are unclear at times, and by returning to the original material we believe that is is now possible to get the real experience of the Poeme, with both visual and audio senses. In this paper we describe the state of the original material, and by considering musicological evidence, we are led to a particular style of reconstruction. The VEP project has led to a true multidisciplinary approach, with input from composers, musicologists, performers and computer scientists., The Poeme Electronique represents an unicuum in Varese's production, being the only pure electroacoustic work the French-American composer ever realised. French acousmatic composer Frangois Bayle is one of the few musi1 http: //www. and funded by EUCulture2000 Andrea Valle, Vincenzo Lombardo Virtual Reality and Multimedia Park, and University of Turin cologists to consider the Poeme electronique as one of "trois geste-titres" of Varese together with lonisation and Deserts[1]. While recognising the primary importance of the Poeme within Varese's output as a whole, he argues that it is a "geste isole" in Varese's career. This issue of isolation among the not-so-large Varesian corpus of works has lead to an analogous isolation in terms of analytical studies dedicated to it [7, 12, 9, 4]. This circumstance is due to the lack of two features. First, the absence of acoustic instruments does not allow an immediate comparison with the sonic world evoked in other pieces, at least in terms of a familiar instrumental landscape implied by sound[2 1]. Second, the consequent absence of a written score for the piece seems to mean we cannot use the analytical categories developed for the traditionally scored works. So, the studies dedicated to the composer's works as a whole assume, implicitly or explicitly, that the Poeme Electronique is not very relevant in the context of the rest of the Varesian music: for Manzoni "the only electronic composition" by Varese "cannot be considered at the same level of his instrumental works, and does not represent a starting point for a period of new experimentations" (preface to [17]) In other cases the scholars claim that a unified approach capable of giving it a precise place among the composer's scant output is impossible. As a relevant example, Bernard[2] (the most influent, complete, documented, and actually sole text dedicated to the works of Varese) covers the whole production of the composer, but it excludes a priori the Poeme Electronique and the other short electronic piece, la Procession des Verges, on the grounds that "without exact notation to work from, not meaningful analysis is possible" (page 243 note 34). This radical decision forbids Bernard even to apply the set of operations on pitch space he himself has developed starting from the emphasis in Varese's texts on space and on spatial transformations of music materials (a poetics which can be traced back in Cubism, Futurism, and more generally in the non musical Avant-gardes of the first years of the 20th Century). We must therefore turn to Cogan's categories[6] for the analysis of the Poeme Electronique. In his later work [7] Cogan addresses the Poeme Electronique via the usual method of investigation through sono

Page  00000002 grams. In this way the "oppositional poetics" of Var~se is revealed by the presence of "highly contrasting regions of audible space (from the most grave bells to the most acute sirens), highly contrasting successions of action (frYom sustained tolling to clipped tapping), highly contrasting sizes and types of sound spectrum (from narrow to wide; and from multiple harmonic strands to dense noise-bands), as well as highly contrasting types of motion (from level stasis to parabolic fr-ee-fall)" (-page 33). Even if these features, working both at a global and at a local level, can represent a sonic fingerprint, which is recognisable in the whole Vari~sian corpus, the way in which these same features have been implemented in the sonic realisation of the Po~me remains mysterious. This is not only a historical problem but also an issue concerning a phenomenology of compositional practices. We now consider the evidence of the remaining original materials, and especially the tapes. 2. HISTORICAL BAC KGROUND When in November 1960 the studio for electronic music (STEM) at the Plompetorengracht in Utrecht (later to become the Institute of Sonology) was founded, not only the equipment from the Philips Studio in Eindhoven was brought in, but also a small archive with tapes of the pieces that were realised there in the past five years. This archive included the tapes of the Po~me Electronique that Edgard Vari~se had composed for the Philips Pavilion at the Brussels World fair in 1958. Since there is some evidence especially from the sketch in figure 2 of the existence of different tracks, we present the knowledge here as a best-guess, that is then confirmed by the investigation below. As was general practice in electronic music production at the time, Var~se composed his music on separate mono tapes (3oips) that had to be played simultaneously in order to hear the complete work (three in this particular case). Some sounds from tracks 2 and 3 were later removed from these tapes and treated with panning and reverberation effects, the results being recorded in stereo on a fourth tape. The now five tape channels were then played on four machines while being copied to a three track 35mm perfo tape (the left channel of the stereo tape being mixed with mono tape 2, and the right channel being mixed with tape 3) that was used during the performance in the Pavilion. During the performances the three tracks were automatically routed through a system of more than 300 loudspeakers (most of them were satellites on the walls of the pavilion and 25 were sub woofers in the foundations). Vari~se took copies of the three original mono tapes with him to the US, where they were synchronised once more by the technicians of Columbia Records for a stereo gramophone record (MS 6146) which has large amounts of added reverberation. In Holland, in the sixties, a 4 track 1 inch tape was -pre pared from the 4 separate tapes at STEM by Frits Weiland, and this was from then on referred to as "the master" of Poi~me Electronique. Most European releases on record or CD were based on this tape, and the original tapes 2 were stored in a safe of the University Utrecht and ignored. STEM, renamed the Institute of Sonology, moved in 1986 to the Royal Conservatory of Den Haag, and a large tape archive was formed that includes all tapes from the Institutes history, including the original Po~me tapes. After investigating these tapes in 2000, Kees Tazelaar realised that with the aid of the computer, a much better synchronisation of the 5 tracks could be made. It was a surprise to discover that the tapes not only contained the master recordings of Var~se's music, but also the 2 minutes of electronic music by lannis Xenakis used as an interlude between performances, which later was published in a slightly longer version as Concret P-H II. An 8 track interpretation of the complete piece was presented at the occasion of the re-unveiling of a sculpture by Le Corbusier (the "Mathematical Object" that stood at the entrance of the pavilion) at the Technical University in Eindhoven in 2000. Since then, a number of different multi-channel versions have been made for various occasions by Tazelaar; for example the new Concertgebouw in Brugge, Belgium, has a 14 channel version as a permanent installation in a special part of the building called the Hanging Rock, which was dedicated to lannis Xenakis by the architect. The Po~me E~lectronique can be heard there at any time at the push of a button. To make it possible to hear the work in its true setting, there is a project in Holland to rebuild the pavilion. The alternative is to create a virtual reality version, which is the aim of the Virtual Electronic Poem project. 3. DETAILS OFi THE AUDIO In order to synchronise the 4 original separate tapes while being played back from individual machines, a sync signal was recorded simultaneously on the four tape recorders and spliced ahead of the Po~me tracks as a lead-in tape. This signal consists of regular metallic clicks and a man's voice, saying: 1, click, 2, click click, 3, click click click etc. When the four machines were started for the final mix-down to perfo tape, these regular clicks enabled the technicians Jan de Bruyn and Anton Buczynski to get the playback in sync before the actual music would start by slightly manipulating the tape flanges manually[3]. For the digital synchronisation, the tapes were not played back simultaneously but one by one on a single machine (a Studer A-8 10 which was optimised for playing back these old 30ips mono full track recordings). Eventually all the material was transferred to the computer and aligned. Roughly, the tapes can be divided into three sections: A: sync section (voice counts, clicks) some minutes B: Concret P-H (3 mono tracks only) 1:52 ~ 0:08 silence C: Po~me Electronique 8:00 When using the clicks to align the tracks, one immedi ately sees that the original machines were not completely 2 now wrongly labelled as being only 'material'

Page  00000003 .y'.... 5:::;**: t. *:^***^^^ ^-^ Figure 1. Sonogram of last 30sec identical in their tape speeds 3. Furthermore, when optimally aligned using the sync part, the ends of the Poeme would not fall together anymore. An attempt has been made to do minuscule transpositions in order to correct for these variations in tape speeds, but with non-satisfactory results. Therefore the Tazelaar reconstruction left the sync signal as it was, and started using other sources as a reference. The ultimate reference (although being made without Varese's presence) would be the 3 track perfo tape from the Sonology archive, which is identical to the version performed in the pavilion. This tape however is in very bad condition and equipment for playing it back is not available. The original film with images from the performance in the pavilion survived too, and for some reason this film contains a mono soundtrack of the Poeme. The Philips Company Archives supplied a copy of this film with the soundtrack. For previous versions of the Poeme Electronique this soundtrack was used as a reference, but at some points from a musical point of view this was not completely satisfying. For the final version it was decided to use the Columbia gramophone record version as a reference, which was authorised by Varese, whereas the Philips one might not have been. Indeed there are some interesting differences; some of them so small that tape speed drifts might have caused them, others so large that only corrections in the edits later on at Columbia could have caused them. Having completed this new synchronisation, we have to make decisions on the spatialisation of the individual sounds. Although the sound routes themselves are well known, both from documentation in Philips Technical Review and from analysing photos, how exactly every sound was supposed to move through which sound route is only documented for 30 seconds of the Poeme (2:05 till 2:35). (Recovering the complete control score for the automated spatialisation is therefore still one of our great wishes). In order to complete the spatialisation we consider that it is necessary to get to understand the work in its time and context. 3 This was a known problem of the so-called Viennese machines, which is why people at the studio for electronic music would usually keep count for which tape was recorded on which machine Figure 2. Varese sketch of last 30sec 4. THE EVIDENCE OF THE CORPUS Has the radical change of medium, from traditional notation and acoustic instruments to magnetic tape and electronic devices, required of the composer an analogous radical departure from technical musical background? If so, an effective gap can be traced before and after the Poeme, at least from this point of view. In this case the resistance of the new medium could have forced the composer to a change in his way of projecting the sound design. This is the assumption of Manzoni, and effectively it can be easily demonstrated that Varese was technically really unskilful with respect to electronics 4. However, from an indication of Vivier, it clearly emerges that Varese developed in composing the Poeme his ideal of a "seismographic notation": Varese n'utilisa pas de port6es musicales inutiles; il dessina des diagrammes sur de larges feuilles qu'il deployait, oui les sons se trouvant indiqu6s seconde par seconde. Sur une tres longue feuille, qui evoquait les anciens manuscripts sur rouleaux, Varese avait note a l'encre rouge quelques definitions, sur la dynamique, les sons transitoires[19] The actually available sketches by Varese (very few, as they share the tragic common fate of the Philips Pavilion and of the other related documentation[15] have always been considered nothing more than quite preliminary drafts, with approximate indications and occasional comments, far from being considered relevant as the instrumental scores. However, a deeper analysis demonstrates, at least for some passages, that the way Varese 4 Varese explicitly wrote to Vivier: "le probleme electronique est plus complexe qu'on le pense, ou qu'en ont conscience Paris, Cologne, et autres colonies plus or moins 'concretes' "[19]. The technical description he gave to Schuller[11] is quite imprecise. This aspects emerges also from the technical description by Tak[13] of the work he made with Varese in the Philips Electronic Music Studio.

Page  00000004 Figure 3. Density 21.5 organised electronic sounds in the Poeme was the same he used to organised instrumental sounds in his other works. Thus the sketches gain a status of accurate formalisation of sound material, which puts them at the same level of musical notation; hence it is possible to trace the continuity of Varese's way of organising sound material. In particular, with respect to Bernard, the whole set of spatial operations developed to analyse the varesian corpus can be considered relevant also for the Poeme Electronique. For our reconstruction the implication of this analysis is that we must follow the orchestral and instrumental music of Varese to guide us when the extant documentation is lacking in precision. 5. COMPUTER INVESTIGATIONS A comparison between a sonographic analysis of the final seconds of the Poeme (figure 1) and one of the surviving sketches from Varese's own hand (published by Treib; figure 2), and in conjunction with careful listening, reveals a high degree of correlation. The rhythm of the notated sequence at the bottom of the sketch can be easily followed by the ear; the semi-pitched clicks match the notated rhythm of the notes, while the low crashes (visible in the sonogram as low-frequency bursts) match the timing of the thick vertical arrows. The wavy line at the bottom of the sketch (indicated with the circled numeral 3) clearly corresponds to the dense jet sound dominating the low-frequency regions of the sonogram. Finally, and most easily apparent to the eye, note the collection of broadband pitched sweeps that accumulate towards the end. The track marked (1) in the sketch clearly matches the loudest trace on the sonogram, while the lowest track on the sonogram matches the line in the upper half of the sketch marked with periodic crosses. We can deduce from this that these pitch sweeps were not casually drawn, in the manner of an improvisatory performance of the electronic instruments; rather the timings and ambitus of the various sweeps were carefully designed in a manner suggestive not of a proto-Cagean indeterminacy but of classic counterpoint. A valuable insight is also provided by the curious circular quadrant diagram at the lower left of the sketch, which is annotated "cumulative amplification". This indicates, especially given the copious dynamic markings, that the effect sought was that of a final gesture of "shock and awe", as the sounds were routed cumulatively to more and more speakers within the space, giving what must have been an overwhelming immersive experience to the listeners. It is important here to remember that not only were artificial sounds of this kind novel to most people, but so also was the sound of the jet aeroplane- the first regular transatlantic service began also in 1958. People flocked in large numbers to airports such as Heathrow simply to watch the arrival and departure of planes, at intervals somewhat greater than we know today. A further insight into technical aspects of the production and performance of the Poeme is seen in the tape edit clearly visible in the sonogram, taken from the Columbia recording, at the point where the pitch sweeps flatten out to their final pitches. This edit is especially apparent in the left channel (though clearly present in both - note that not only the pitch sweeps are affected by this edit; so also is the general background noise). It is clear from the sonogram that one pitch sweep is unaffected by this edit, indicating clearly that this sound was on a tape that required no editing. We speculate that one or more tapes may have suffered from slight stretching or variable machine speeds, and were consequently shortened in this way to ensure all tapes ended exactly together. Finally, we note the various commentaries that suggest that the Poeme, for lack of a formal score, cannot be approached by musicologists as they would approach any of Varese's instrumental works. The implication is that in some way the Poeme represents some departure or digression from his established compositional methods. We do not agree with this view. Similar gestures to those found in the Poeme can be found throughout Varese's work, and not only within those calling on massed instrumental forces. We quote the final phrase of Density 21.5 for solo flute, composed in 1936 at the commission of the flautist Georges Barrere, for the inauguration of his new platinum flute (figure 3). In this bravura gesture the music sweeps (already fortissimo) via the low C up almost three octaves to a high B, at which point the player is called to project with still greater (seemingly limitless) force. This gesture compares very closely to the closing seconds of the Poeme. We can well imagine that Varese saw the progress of technology in almost mythic terms; the shock and awe of the jet engine and electronic sounds reflect one aspect of the modern age, but the huge crescendo finale from the new platinum flute reflects another, so it cannot be surprising that Varese should celebrate them both with gestures whose extraordinary similarity transcends any disparity of medium. Varese did after all say Avec quelle nature etes vous en contact? Moi, je parle de la ville o*ij'ai les plus vecu: New York. A New York, il y a certainement des gosses qui n'ont jamais vu un ruisseau. Qui n'ontjamais entendu le chant des oiseaux. Qui n'ont jamais entendu, je ne sais pas, moi, tout ce qui peut passer dans la campagne. Mais qui sont familiers avec le vrombissement des avions, avec le bruit des autos, avec les sons industriels, avec tout ce qui passe dans une mtropole comme New York. Pour eux c'est peut-6tre ce qui represente les bruits de la na ture: le milieu dans lequel ils vivent et les choses travers lesquelles ils reagissent"[5]. As Bernard notes[2] on page 41 "Spatial music and

Page  00000005 1 -.A -- 'l -,: ---C vs i m"''" * ' rX C 'J - 't:........ d --**^p^ ''~sH^ ^ ' "...... \, SI i Z^?. J-.1 *- -. V ^, \ i^S-~ ft"-^' "^ " ^ ": *'"'':^ ~JI ~,: ~. ~~~~ a' -^felA ^ j.^/-^ Figure 5. Sound Routes music in space are phrases used by Varese over and over again; the single unifying principle of his music is the manipulation of materials with reference to a spatial framework". Within Varese's instrumental works this spatial paradigm is at best implicit, in the juxtapositions, expansions and contractions of pitch, register, timbre and rhythm, offering ample opportunities for both macro- and microcosmic analysis. Bernard himself presents a vivid analytical framework based on these primary geometric processes of projection, rotation, expansion and contraction. A work such as Density can but suggest such dimensionality in a microcosm; the Poeme in contrast demonstrates Varese's attempt to make at least some of these dimensional aspects explicit within the literal spatial framework of the Pavilion. It is deeply ironic that, by all accounts, Varese hardly engaged with the details of the spatialization process, and we can but speculate as to what would have transpired had Varese engaged more directly, and directively, with the formidable and undeniably complex resources available to him. 6. POSITIONAL RECOMPOSITION OF THE POEME For the purposes of the project, the task of the audio reconstruction team could be reduced to that of creating a set of soundfiles corresponding to the sound routed to each speaker; thus some 350 sound files would define the positional rendering of the whole work. Complicating that otherwise purely tedious task was the fact that neither the original control score nor the control tape was available (save the 25-second extract reproduced in [10]). Also, for rendering purposes a listener position must be defined. This meant that a more creative, compositional approach was required, leading in turn to a goal to implement a representation of the speaker distribution that would enable composers to exploit the space with their own sounds. To this end, we chose to implement the sound routes and clus ters as Csound instruments. As supplied, the speaker coordinates took as their origin a position just outside the space (see figure 5). For testing in Csound we have added 10 metres to the Y position to locate a listener approximately in the middle of the space. In recent years Csound has undergone (and indeed continues to undergo) considerable transformation from its origins in the Music 11 system[18]. Real-time interactive performance has now been available for a number of years, in several complementary versions exploiting the facilities of a particular OS. Csound supports multi-channel real-time output (both writing to disk and rendering directly to audio), and recently contributed opcodes include several tools for 3D spatialisation, including a number of Ambisonic opcodes by Varga implementing B-Format[14] 5 encoding and decoding. For full (dynamically calculated) spatialisation these opcodes are very CPU- intensive. One of the Varga opcodes (spat3dt) enables a synthetic B-format impulse response, corresponding to a particular room definition, to be written to a Csound f-table, for use with an FFT-based convolution opcode such as the low-latency fast convolution opcode pconvolve recently contributed by Ingalls[14]. Csound enables alternative approaches to implementing the Poeme to be explored very easily. We illustrate here a simple approach in which the speaker position data for each sound-route or cluster is stored in a set of three ftables containing respectively the X,Y and Z co-ordinates (see figure 4). While the original form of Csound required f-tables to be defined in the score file, recent versions permit them also to be defined in the orchestra file. While not exactly "private" in the OOP sense, this nevertheless enables the score to be less cluttered, and may discourage the composer from the temptation to modify the canonical positions. For the sound-routes one instrument is sufficient, receiving as a score parameter the number of speakers in the route (see figure 6). Not shown in these examples is the case of a single long sound such as the pitch sweep featured several times in the Poeme. The standard Csound solution here is to write a "legato instrument"[8], which effectively "ties" one note to the next, retaining the internal state of the instrument (such as position within a file); such a note is indicated by a negative value for p3 (the duration parameter). The spat3di opcode generates encoded first-order B-Format data; the complete instrument incorporates the option to decode to speaker feeds (e.g. simple rectangular quad for horizontal-only rendering), via the last instrument parameter. Experienced Csound users will easily suggest alternative, more self-contained, implementations; this is an aspect under constant review, as we seek a form both flexible and precise enough to enable composers to exploit to the full the opportunities offered by this unique space, while minimising complexity in the score. SAmbisonic is a registered trademark

Page  00000006 Figure 4. A Poeme sound route (a) in a Csound score f-table route V: 13 speakers f51 0 16 -2 -16.7 -17.15 -16.2 -15.9 -15.3 -15.3 -14.9 -13.9 -14.2 -13.5 -13.3 -17.15 -17.0 f52 0 16 -2 7.7 6.95 8.5 9.6 10.7 12.1 13.1 13.5 14.45 13.6 13.2 6.267 5.516 f53 0 16 -2 10.25 11.3 9.15 7.9 6.45 5.4 4.1 3 2.3 2.2 2.3 12.42 13.321 (b) defined in the orchestra ir5l ftgen 1000, 0, 16, -2, -16.7, -17.1 -14.9, -13.9, -14.2, -13.5, -13 ir52 ftgen 1000, 0,16, -2, 7.7, 6.95, 8. 14.45, 13.6, 13.2, 6.267, 5.516 ir53 ftgen 1000, 0, 16, -2, 10.25, 11.3, 2.3, 2.2, 2.3, 1, 2.42, 13.321 7. CONCLUSIONS We conclude that using evidence from the tapes, from the corpus of Varese's work and the use of computers we can reconstruct the spatialisation of this important work, at least such that it is in keeping with the composer's wishes and style. The article by Varese[16] contains excerpts from lectures by Varese, compiled and edited with footnotes by Chou Wen-chung, the disciple of Varese who also completed the unfinished Nocturnal. As Varese never officially authored a book, quotations of his lectures are spasmodic and never in a complete form. Even if the article is a "considerable abridgement" of the Schwartz and Childs anthology, the excerpts contain some of the most famous statements by Varese about sound and space, music composition as sound "crystallization" and "organized sound". He also explains his beloved quotation by H6ene Wronsky. The first excerpt, "New instruments and new music" from the famous 1936 lecture in Santa Fe, seems to represent the most relevant declaration of the Varesian poetics of the space (to quote Bachelard), in which he speaks of collision of("sound masses", and of planes "projected onto other planes". This spatial metaphor is intended by Varese to be acoustically realised in a physical space, thus becoming discernible to the listener, and permitting "the delimitation of [...] Zone of Intensities. These zones would be differentiated by various timbres or colors and different loudnesses". Varese strives for the "sensation of nonblending" between adjacent zones. Moreover, "new notation will probably be seismographic". Thus, the Poeme seems to be the effective realisation of Varese's dream, even if at that point, he probably did not know how technically to face the situation (thus demanding from Willem Tak, the Philips' technician devoted to the audio installation for the pavilion, the whole work of intonation). Thus we have based this reconstruction on the triple pillars of the published fragments, the poetics declaration by Varese[16], and the evidence of the other works. Wen-chung, a dedicated disciple of Varese, discusses 5, -16.2, -15.9, -15.3, -15.3, \.3, -17.15, -17.0 5, 9.6, 10.7, 12.1, 13.1, 13.5, \ 9.15, 7.9, 6.45, 5.4, 4.1, 3, \ some themes of Varese aesthetics and music production. In particular, he observes that some of the ideas developed for the project of Espace for orchestra and chorus, aborted because of the lack of the necessary electronic equipment, "finally found their way into Deserts and Poeme Electronique"[20]. 8. ACKNOWLEDGEMENTS This project was funded by the EU Commission in the Culture 2000 programme (ref. No. CLT2004/Al/CH/IT352). In addition Andrea Valle benefits from a fellowship granted by Fondazione ISI. Kees Tazelaar is at the Institute of Sonology/Royal Dutch Conservatory, and is a Visiting Research Fellow at the Department of Computer Science, University of Bath. John ffitch and Richard Dobson are also members of the Media Technology Research Centre at University of Bath. 9. REFERENCES [1] Frangois Bayle. Musique acousmatique, propositions... positions. INA-Buchet/Chastel, Paris, 1993. [2] Jonathan Bernard. The music ofEdgar Varese. Yale UP, 1987. [3] Anton Buczynski. Private communication with K. Tazelaar, 2004. [4] Densil Cabrera. Sound Space and Edgard Varese's Poeme electronique. Master's thesis, University of Technology, Sydney, 1994. [5] George Charbonnier. Entretiens avec Varese. Belfons, Paris, 1970. [6] Robert Cogan. New Images ofMusical Sound,. Haryard UP, 1984. [7] Robert Cogan. Varese, An Oppositional Sonic Poetics. Sonus, 11(2):26-35, 1991.

Page  00000007 [8] Richard Dobson. The Csound Book: Tutorials in Software Synthesis and Sound Design, chapter 7: Designing Legato Instruments in Csound. MIT Press, February 2000. [9] Lawrence Ferrara. Phenomenology as a Tool for Musical Analysis. The M2usical Quarterly, KXX(3):355-373, 1984. [10] Jean Petit. Le Po~me Elctroniquke. Le Corbusier, Paris, 1958. [1 1] Gunther Schuller. Conversation with Var~se. In Benjamin Boretz and Edward T. Cone, editors, Perspectives on American Composers, pages 34-39. Norton, 1971. [12] Ann Stimson. Analyzing Po~me 6lectronique: Clues from the Getty Archive and the Spectrogram. Contempzorary M~usic Forum, 3:12-26, 1991. Proceedings of the Bowling Green State University New Music and Art Festival. [13] Willem Tak. Les effets sonores. In Le Pavillon Philips d 1 'Exposition U/niverselle d~e Brukxelles 1958, volume 20, pages 1-3. Revue Technique Philips, 1958. [14] The Csound Development Team. The Canonical Csound Reference Manual, 2005. http: // c sound/manual!/. [15] Marc Treib. Space calculated in seconds: the Philips Pavilion, Le Corbuksier, Edgar Var~se. Princeton UP, 1996. [16] Edgard Var~se. The Liberation of Sound. In Benjamin Boretz and Edward T. Cone, editors, Perspectives on American Composers, pages 25-33. Norton, 1971. [17] Edgard Vari~se. Ecrits. Bourgois, Paris, 1985. [18] Barry Vercoe. Csound A Manual for the Audio Processing System and Supporting Programs with Tuktorials. Media Lab, M.I.T., 1993. [19] Odile Vivier. Var~se. Seuil, Paris, 1973. [20] Chou Wen-chung. Open rather than bounded. In Benjamin Boretz and Edward T. Cone, editors, Perspectives on American Composers, pages 49-54. Norton, 1971. [21] Trevor Wishart. Sound Symbols and Landscapes. In S. Emmerson, editor, The Language of Electroacoukstic 1Music. MacMillan, London, 1986.

Page  00000008 Figure 6. Sound route example. (a) Instrument definition instr 3 iftx = p5 iy = 10; approx middle inside ibf = p8 ifty = iftx + 1 iftz = ifty + 1 ispkl = p7 ispk2 = ispkl + 1 ispk3 = ispk2 + 1 ispk4 = ispk3 + 1 ispk5 = ispk4 + 1;params for spat idist = 1 ift = 0; free field; can also use f-table for room specification (slower) imode = 3; output wxyz signals asig soundin p4 asig = asig * p5;speaker 1 ixi table ispkl,iftx iyl table ispkl,ifty izi table ispkl,iftz; and so on for speakers 2 -5 then spatialize each speaker to B-Format awl,axl,ayl,azl spat3di asig, ixi, iyl - iy, aw2,ax2,ay2,az2 spat3di asig, ix2, iy2 - iy, aw3,ax3,ay3,az3 spat3di asig, ix3, iy3 - iy, aw4,ax4,ay4,az4 spat3di asig, ix4, iy4 - iy, aw5,ax5,ay5,az5 spat3di asig, ix5, iy5 - iy,; mix and output; ifac = 1; (adjust ad lib: most sounds v aw = ifac * (awl + aw2 + aw3 +aw4 +aw5) ax = ifac * (axi + ax2 + ax3 +ax4 +ax5) ay = ifac * (ayl + ay2 + ay3 +ay4 +ay5) az = ifac * (azi + az2 + az3 +az4 +az5) izl, iz2, iz3, iz4, iz5, idist, idist, idist, idist, idist, ift, ift, ift, ift, ift, imode imode imode imode imode low level) outq endin aw,ax,ay,az (b) Note statement from score: each line advances one speaker;p8 sets decoding: 0 = decode, i3 0 0.25 "blup.aif" i3 +. "blup.aif" i3 +. "blup.aif" i3 +. "blup.aif" 1 = output B-format 11 51 0 0 11 51 1. 11 51 2. 11 51 3.