HighC est inspiré de l'UPIC de Iannis Xenakis, développé dans les années 80. A la différence d'autres systèmes de "peinture de son", HighC utilise un modèle de synthèse additive structurée. Ce modèle permet la composition d'objets sonores simples en objets de plus haut niveau qui peuvent être manipulés comme des symboles. En gros, HighC est à HyperUpic ou MetaSynth ce que Powerpoint est à Photoshop.

La synthèse sonore "graphique" et l'upic

Iannis Xenakis (1922-2001) était un architecte et un compositeur qui imagina un moyen visuel de créer de la musique dans les années 70. Son système, l'UPIC (pour unité polyagogique informatique du CEMAMU), utilisait une tablette graphique et un ordinateur très couteux.

Le concept de base de l'UPIC, ce que j'appelle la synthèse sonore "graphique", est à la fois simple et très expressif. Une composition se présente comme un ensemble de courbes continues sur une partition, dont la progression horizontale détermine l'évolution de hauteurs en fonction du temps. D'une certaine façon, c'est une généralisation de la notation traditionnelle de la musique occidentale, qui rend la partition linéaire et continue dans les domaines temporel et harmonique. Une illustration vaut plus que toute explication textuelle: voici l'une des premières pièces créées avec l'UPIC, Mycenae Alpha:

(Traduction a continuer...)

The beauty of graphical audio synthesis is that it is usable by the most novice musicians, even children, and still it provides a huge expressive space for the most creative and demanding composers. Anybody can start creating their own compositions in HighC or UPIC in a matter of seconds. Confirmed musicians are offered new ways to explore soundscapes, literally to "create the unheard", which is the essence of all artistic endeavor.

Yet, below the apparent simplicity of the concept lie a number of issues and compromises to ponder over. The implementation of the synthesis technique is quite trivial for anyone knowledgeable in digital audio processing. The first issue I call the "representation" problem: First of all, a simple sound is not simply a trajectory of pitch in time. A sound has other evolving attributes, such as its intensity and timbre. For a faithful representation to be provided, those attributes should be encoded in the visual image of the soundscape. Second, because our visual and auditory systems have very different ways of composing elementary signals to give them a meaning, any visual representation of audio is doomed to miss something of the audio stimulus, or to create a perceptual distortion. As an example, I believe the main weakness of the current approaches is that they do not reflect the harmonic properties of sound: 2 sounds separated by approximately an octave look quite the same on the score, but they sound very differently depending on whether they are at unison or separated by a 7th or 9th.

Still, UPIC encountered much enthusiasm and was used by numerous contemporary music composers in the 80's and 90's despite its limited availability. Despite newer versions being written, the software-only UPIC has not been widely distributed and is not maintained anymore. Still, the CCMIX continues to maintain and evolve Iannis Xenakis' musical vision nowadays. Fortunately, thanks to nowaday's progress in computing power, massive additive synthesis, at the heart of UPIC and HighC synthesis models, is now available to a large audience on commodity hardware, with far more features and ease of use.

Synthèse sonore graphique "organique" ou "symbolique" ?

UPIC has inspired quite a few other interesting tools, such as HyperUpic and its heirs: Metasynth and Coagula (or try at Unesco). Yet, somehow, the most successful of those inheritors are bitmap-based rather than vector-based as was the original UPIC: the image is interpreted as a sonogram which is turned into a sampled sound by means of some kind of inverse Fourier transform. This synthesis technique provides very good means to literally chisel audio objects, add or remove effects at a very fine granularity. This technique definitely has its uses, and all those programs are great tools in the hand of capable musicians or even amateur to "create the unheard".

One of the key specificities of HighC over a few of these counterparts is that it is a symbolic system rather than an organic synthesis system. In HighC (as in UPIC), a piece is made of sound curves, and each curve is an individual object with its own properties, all numerically described. The composition is not an organic bitmap from which the acoustic phenomena emerge through a unique transformation. Simply said, HighC is to HyperUpic what MS PowerPoint is to Adobe Photoshop, or what Meccano is to Plasticine.

This has one drawback, namely that the representation of the auditory experience is not completely truthful ; it leaves room for interpretation. For instance, a sound with a noise waveform will be heard exactly the same whatever the pitch at which you place it.

However, I believe it is interesting to keep the ability to manipulate sounds as symbols: it lets you introduce the power of language (or assembly of symbols organized by rules) into your music creation activity. By "power of language", I mean that HighC makes it possible to create and manipulate higher level abstractions made of low level sound components, and combine them with each other transparently into even more sophisticated compositions.

It turns out, it is the essence of music composition to create such abstractions from raw acoustic events and combine them according to one's whim into more complex structures. Whether or not music is a language is a debate I won't enter into. In any case, HighC lets you create a language to manipulate sound, rather than propose you a single method for chiseling into a glob of sound.

The goal of HighC is to integrate this musical-language-building approach into a graphical audio composition tool.

L'approche de highc sur les problemes de representation et de structuration de la partition

I was lucky enough to do a month-long internship in 1984 as a high school student to experiment with UPIC. The concepts behind this instrument, most notably its ability to create a sense of synesthesia have kept fascinating me, as I think they reveal a lot of insights on what human computer interaction is about: transforming objects made of bits into representations made of atoms and vice-versa.

My own specialty is not music, but human-computer interaction, or, as I'd rather call it, interactive information systems design. The area of audio-painting/drawing raises numerous challenges, issues and compromises about representation, structuring and manipulation vocabularies, where I can put my experience at play.

-

HighC represents my solution to the representational and structuring dilemma I mention above. The software is still far from what I envision currently, and my vision still needs to be elaborated and explicated. A newer version of this page should fill the gap to explain the choices made here.

-

I started working on my own sound synthesis algorithms in 1991, had an extensive work phase in 1998 to polish up the model. I finally started writing a user interface to create music by drawing it in 2006. Version 1 was released in March 2007, and presented at IRCAM in June. In November 2007, version 1 had reached 10000 downloads, a sure mark that some interest is being raised. Still, the present version is only a sketch: it needs a lot of polishing and user testing to become a real instrument, and there are many more features that I intend to implement over the next few years to provide better control of the sounds, their dynamics, modulation and spatialization.

In releasing a partially featured version, my intend is to gather useful feedback early on, rather than deliver too complex a software that would not satisfy important requirements.

If you know how to use a drawing program, you'll know how to use HighC. I wish readers to try out HighC, and report on their uses on the forums.

Last, but not least, HighC is pronounced with a strong French accent: "I see!".

 

PS: I should add a few references here. I recommend reading Pierre Couprie's work, among which: "Graphical representation: an analytical and publication tool for electroacoustic music" or in French: La musique électroacoustique : analyse morphologique et représentation analytique.