Musical Structures and Neural Networks: From Synchronous Polyphony to Algorithmic Ecologies

Abstract

This study explores the continuity between musical structures and neural networks, framing creativity as an oscillation between rules and emergence. It examines how generative systems—mechanical or digital—encode style through combinatorial or probabilistic grammars, from eighteenth-century dice games and Kircher’s Arca musarithmica to convolutional and deep learning models. Using sources such as Musurgia Universalis, Traité de l’harmonie, and Opticks, alongside computational reconstructions, the analysis highlights structural analogies across epochs: stratification, proportio,n and algorithmic logic as shared principles of art and computation. The conclusion argues that algorithmic ecologies redistribute authorship, turning composition into a dialogue between human intention and machine inference. In this technocultural context, music and AI converge as architectures of meaning.