Agostino Di Scipio

In the mid-1970s Xenakis began developing the UPIC system, a computer music system having a drawing surface as input device: drawings made on a CAD-like tablet (Computer Assisted Design) could be made into sounds through a software interface, a dedicated processor and a digital-to-analogue converter. UPIC (Unité Polyagogique Informatique du CEMAMu [Centre d’Études de Mathématique et Automatique Musicales]) represented a more intuitive and interactive approach on digital sound synthesis than Xenakis had outlined earlier on, with his stochastic synthesis (a whole set of direct sound synthesis algorithms based on stochastics and probability theory) (Xenakis 1992, 242-254.

Although sometimes described as “unprecedented,” UPIC was actually not the very first case of sound synthesis with graphical methods. One may recall various earlier efforts, including Percy Grainger’s Free Music Machine (electro-mechanical devices, late 1940s), Osmond Kendall’s Compositron (1950s), or Fernando von Reichenbach’s Convertidor Gráfico Analógico (the latter was built at the Laboratorio de Musica Electronica in Buenos Aires in the late 1960s, and allowed crude but efficient screen-based controls of Moog synthesizer modules). Computer music developments sharing such an approach had been pursued in North America (at the University of Ottawa and at Bell Telephone Labs, New Jersey) not long before Xenakis took up the UPIC project. This list of historical precedents would further expand should we also include methods of optical synthesis (on film) developed in the Soviet Union and elsewhere, even before Second World War.

None of these pioneering projects, however, should obscure the peculiarities of the UPIC system and the large interest it raised as a creative tool. The unusual approach undertaken with the UPIC seemed after all very apt for a composer putting both music and architecture at the focus of his creative practice. In the context of music educational activities supported by the CEMAMu, Xenakis regularly invited young people and non-experts to play with the UPIC as a way to get acquainted with notions of acoustics and music (Nelson 1997; Deprés 2020).

The first complete UPIC prototype dates back to 1977 and was the outcome of efforts begun a couple of years before together with Patrick Saint-Jean, a young engineer interested in the potential that the then-new mini-computers had opened up for man-machine interaction (Saint-Jean 1977). The UPIC prototype worked with a Solar 16-40 mini-computer. Although the sound synthesis process initially was still non-real-time, at least the graphical interface allowed for a more direct, interactive way to prepare and handle input data. At that time, the CEMAMu technical team also included Cornelia Colyer and Guy Médigue, but significant UPIC developments took place slightly later, as they were taken up by Gérard Marino, Jean-Michel Raczinski and others (Marino et al. 1993).

Xenakis started exploring the UPIC musical potential with Mycénes Alpha (1978), his first work entirely made of computer-generated sounds (editing and mixing were still carried out by the conventional analogue means). Itwas initially conceived to be part of the multimedia show Polytope de Mycénes (1978) and would be later presented as an autonomous tape music work. More UPIC versions would follow. The first UPIC that was capable of real-time synthesis dates from 1987 (Raczinski &Marino 1988). Although Xenakis wanted to use it in live performance situations, things would actually turn out trickier than expected. After the problematic première of Taurhiphanie (live in the ancient Roman Arena of Arles, in 1988), Xenakis had to reconsider his plans: in later occasions Taurhiphanie would be presented as a regular tape music work. Let us also remind that, at several moments in the course of the 1980s, Xenakis made plans for a Ballet des robots émancipés, quite a futuristic project with robotic arms that were supposed to make drawings on the UPIC and thus “play” with it. In the late 1980s, Xenakis planned to adapt a few robotic units of the kind then provided by the COMAU research centre to the FIAT car assembly lines, in Turin (apparently, later contacts were made with Renault too). Anyway, none of these contacts would eventually work out. Eventually, the project had to be abandoned, costs being too high. For more on the aborted Ballet see Lohner 2020.

In the meantime, in 1986 Xenakis launched Les Ateliers UPIC, a workshop facility located in the suburbs of Paris, hosting compositional and educational activities centred around the UPIC system. In the 2000s the facility would expand and become CCMIX (Center Création Musicale Iannis Xenakis). The latter would remain in service only until 2007; yet, many younger and more established composers had the chance to work there with the UPIC. In still later years, a number of researchers and musicians would reshape the UPIC entirely in software and had their UPIC derivations circulate either under copyright or as public domain software (Baudel 2006; Georgaki 2015; Scordato 2020). The most recently available implementations even run on mobile devices (Bourotte 2022).

Sound synthesis in Xenakis’s UPIC is basically equivalent to a table lookup digital oscillator: first, a sequence of unit samples (integer numbers, representing the signal waveform) is loaded into a table (a memory buffer, a wavetable); then the program starts picking up samples from there, looping around the wavetable again and again (for brevity, we omit important technical details). The sequence of output samples represents a digital audio signal, which is finally sent through a digital-to-analogue converter to one or more loudspeakers. That constitutes the most basic and widely used sound synthesis process. The peculiarity of UPIC lies in the fact that the wavetable samples are obtained by drawing straight or curved lines on the graphic pad, by hand. In later versions, built after the mid-1980s, the wavetable could also be created by sampling a microphone signal via analogue-to-digital converters. In the premiere of the afore-mentioned Taurhiphanie (1988), Xenakis tried to live-sample the bellowing of real bulls!

A unique aspect of UPIC is that, once a line or curve has been drawn, nothing is really decided about its interpretation as a signal: it can be used as a wavetable in the audio level process, as previously mentioned, but it can also be used as a control signal. In the former case it affects the timbre or sound colour, in the latter, it works as a low frequency oscillator (LFO) – driving the amplitude (envelope shaping) or the frequency (fixed pitch or glissando) in the sound being synthesized. Further, it can even be used as a high-rate modulating signal (amplitude and frequency modulation). This all amounts to saying that the length of drawn lines is not equal to an absolute duration: one UPIC “page” can be assigned a duration of few milliseconds to several minutes, as it is only given an actual duration when getting turned into sound (in early implementations, the range was 6 milliseconds to 12 minutes (Xenakis 1992, 332). In other words, drawings can map into either micro-time or macro-time aspects in the sounding output. Therefore, while UPIC favors the crafting and organization of individually defined sound objects – and that greatly differs from Xenakis’s approach to stochastic synthesis (early 1970s) as well as from the “clouds of sonic quanta” of his prototypical granular synthesis (1959) – it also allows to handle sound objects across various time scales. In a way, the UPIC composer moves or shifts across different temporal layers in the musical fabric.

The particular approach of UPIC inevitably requires a trial-and-error approach. It demands of a composer to double-check that the signs s/he draws may eventually make musical sense as sounds actually heard: the identity of linear drawings (on the bidimensional space of the input pad) and the sound eventually heard (as a phenomenon unfolding in time) is of purely formal order, and certainly does not guarantee against arbitrary mappings. Accordingly, we can say that – just like it had happened with Xenakis’s earlier sound synthesis efforts – the sonic outcomes of UPIC are epiphenomenal: they are indeed side-effects no more of a stochastic process but of a gesture in space made audible by a technical artefact. Exactly there lies a significant and pedagogical potential (Nelson 1997), as in fact the musician is invited to creatively explore a sensible nexus between composing the sounds themselves (micro-time) and composing with the sounds (macro-time).

It seems correct to say that, within the complete picture of Iannis Xenakis’s long-standing commitment to electroacoustic and computer music (Di Scipio & Solomos 2023), the UPIC project stands out as a truly unique endeavour. It provided an opportunity for the composer to reconsider some of his earlier conceptual and technical orientations, but also overcoming some of the difficulties the latter had engendered along the way. And still, upon listening, the music Xenakis has composed with the UPIC relates with substantial continuity to his other electroacoustic (and even orchestral) music. That shines through clearly as we listen to Mycénes Alpha or Voyage des Unari vers Andromède (1989). Their sounding reality is hardly less rough and scathing than the synthetic buzzes scattered around the sonic fabric of La Légend d’Eer (1977) or the biting, noisy textures of Bohor (1962): it unmistakably belongs to the varied sound-world marked by musical works as different as Metastaseis (1954) and Jonchaies (1977), not to mention the fully automated computer music of Gendy3 (1992).


This text is adapted from Agostino Di Scipio, “La Sintesi del suono in Iannis Xenakis. Indagine di una ricerca compositiva,” Musica/Tecnologia, 16, 2022.


Baudel, Thomas. 2006. “From information visualization to direct manipulation: extending a generic visualization framework for the interactive editing of large datasets”. Proceedings of the 19th annual ACM symposium on User interface software and technology.

Bourotte, Rodolphe. 2022. UPISketch; User manual.

Di Scipio, Agostino & Makis Solomos. 2023. La musique électroacoustique de Xenakis, Durand-Salabert-Eschig, Paris (to appear).

Deprés, Alain. 2020. “The UPIC: towards a pedagogy of creativity”, in From Xenakis’s UPIC to Graphic Notation Today (Peter Weibel, Ludger Brümmer and Sharon Kanach, eds.), Hatje Cantz Verlag, 2020.

Georgaki, Anastasia. 2015. “Sound pedagogy through polyagogy”, in Iannis Xenakis, la musique électroacoustique (Makis Solomos, ed.), L’Harmattan.

Lohner, Henning. 2020. “Iannis Xenakis, Robots, and the UPIC – the greatest robot show the world never saw”, in From Xenakis’s UPIC to Graphic Notation Today, edited by Peter Weibel, Ludger Brümmer, and Sharon Kanach. Hatje Cantz Verlag.

Marino, Gérard, Marie-Hélène Serra, & Jean-Michel Raczinski. 1993. “The UPIC system: origins and innovations”, Perspectives of New Music, 31(1).

Nelson, Peter. 1997. “The UPIC system as an instrument of learning”, Organized Sound, 2(1).

Raczinski, Jean-Michel & Gérard Marino. 1988. “A real time synthesis unit”, Proceedings of the International Computer Music Conference.

Saint-Jean, Patrick. 1977. Conception d’un système informatique de laboratoire opérationnel pour le Centre d’Études de Mathématique et Automatique Musicale, Mémoire de diplôme d’ingénieur, École Supérieure d’Informatique, d’Électronique et d’Automatique, Paris.

Scordato, Julian. 2020. “Novel perspectives for graphic notation in Iannix”, in From Xenakis’s UPIC to Graphic Notation Today (Peter Weibel, Ludger Brümmer and Sharon Kanach, eds.), Hatje Cantz Verlag.

Xenakis, Iannis. 1992. Formalized Music. Thought and Mathematics in Music, Pendragon Press. Revised and expanded revision of Formalized Music. Thought and Mathematics in Music, Indiana University Press, 1971.

How to cite

DI SCIPIO, Agostino. 2023. “UPIC.” In A Xenakis Dictionary, edited by Dimitris Exarchos.