From mimesis to katharsis: mathematical formalization and machine learning in music creation

Chair: Xavier Leroy

Abstract

This lecture offers a reflection on the growing role of machine learning in contemporary music creation, at the intersection of artistic intuition and mathematical formalization. Drawing on my work in computational creativity, assisted orchestration, and human-machine co-composition, I will examine how algorithmic systems can move beyond imitation (mimesis) to become agents of sensorial and cathartic transformation (katharsis).

I will explore how mathematical models—far from serving merely as tools for optimization—can structure aesthetic processes, reveal emergent forms, and open new spaces for listening and invention. The talk will highlight systems where neural architectures engage deeply with musical logics, and where symbolic and perceptual representations are brought together in a unified creative gesture.

This epistemological shift, which entrusts aspects of judgment or style to computational agents, challenges conventional notions of authorship, work, and listening. Situated between science and artistic practice, I will advocate for a vision of co-evolution between humans and machines—grounded in shared structures, emergent languages, and the unfinished power of musical creation.

Carmine-Emanuele Cella

Carmine-Emanuele Cella is a composer, applied mathematician, and Associate Professor of Music at the University of California, Berkeley, where he co-directs the Center for New Music and Audio Technologies (CNMAT). He is also Director of Artistic Research at IRCAM (Paris) under a Sorbonne Université chair. His work explores the intersection of advanced mathematics, artificial intelligence, and artistic creation, with a particular focus on orchestration, computational creativity, and hybrid research frameworks that bring together composers, performers, and scientists.