Critical Climate Machine 3/4 : Composing with Dyci2

Artistic Residencies: The Blog

Controlled by Gaëtan Robillard and Dionysios Papanikolaou, the generation engine OM-DYCI2—developed by Jérôme Nika, a member of the Musical Representations team—was fed with 34 dialogues, made up of climatosceptic arguments and their refutations to produce hours and hours of sounds. This plethora of material was then entrusted to Tony Houziaux, a sound designer trained in the world of entertainment and animated film, who is also an electroacoustic improviser and teacher at the Gustave Eiffel University (Marne la Vallée) and at the École supérieure d'art de La Réunion.

“For the Critical Climate Machine, my work focused on two distinct yet related aspects," says Tony Houziaux, "the sound design of the installation (the positioning and layout of the loudspeakers, the occupation of the sound space) on the one hand, and the editing and composition of the raw material produced by OM-DYCI2 (and therefore its spatialization) on the other.

“The idea," says Gaëtan Robillard, "is to use this material as an instrument to fill up the space. When we started work, no one knew exactly what OM-DYCI2 would produce! In the end, we ended up with over 2,000 sound files, weighing in at over 6.5 GB.”

“I listened to everything," says Tony Houziaux. “However, even if I’m not used to getting my fingers dirty and playing with the settings of devices as complex as OM-DYCI2, Gaëtan, Dionysios and I agreed on reference points, degrees of transformation of the material, which allowed us to devise a classification in this mass of sound. A second line of work to question this rich material, and better organise it in time, was sound density.”

The last element on which Tony Houziaux was able to base his composition: the permutation rules that Gaëtan Robillard developed to organise the oppositions between the climatosceptic arguments on the one hand, and their scientific refutation on the other, but also between recorded language and generated language, between typical speech and singular speech.

Interface of the generative engine designed with OM-DYCI2 to generate a large quantity of voices from a collection of "structure" voices and a collection of "texture" voices.

“All of this was already sketching out a compositional space," says Tony Houziaux. Insofar as all these sounds are vocal, without really being voice, but respecting a form of prosody, I wanted the visitor to be able to immerse themselves in them, to be free to make their own journey and form their own opinion when faced with particular distributions of sound in space (obtained by oppositions or permutations). I wanted them in turn to participate in the construction of listening ethics. For example, they may switch without warning from a point of view where preference is given to the intelligibility of dialogue to another where they pay more attention to the sound pattern.