All0Sound

Music can be understood as a system of aesthetic rules that organize sonic events and their transformations over time. AI music, automatic music, and generative music all belong to a broad field of computational approaches for creating musical material. Within this field, two major paradigms are especially important: machine-learning systems, which infer statistical patterns from existing musical data, and rule-based systems, which generate material through explicitly designed algorithms. In both paradigms, human aesthetic decisions remain foundational. In machine-learning approaches, the training data is usually drawn from human-made music, meaning that the model learns from preexisting musical practices, styles, and conventions. In rule-based systems, the composer or designer defines the procedures, constraints, and musical behaviors that shape the generated output. The first paradigm has expanded rapidly with contemporary AI models, but it also raises important questions about style replication, interpolation, authorship, and copyright. The second paradigm foregrounds human agency more directly, since musical style emerges from rules and procedures intentionally designed by a human creator. Algorithmic and procedural systems are especially relevant to videogame audio because their rules can be designed to react, adapt, and interact. Like procedural sound design, rule-based musical systems can respond to gameplay events, player behavior, environmental changes, or expressive parameters. Machine-learning systems can also be interactive through mechanisms such as model weights, prompts, embeddings, or temperature controls, but the musical meaning of these controls is often less transparent from a human aesthetic perspective. A central research challenge, therefore, is how to combine the strengths of both paradigms: the flexibility and pattern-recognition power of machine learning with the legibility, intentionality, and controllability of rule-based design. This search for human-understandable parameters for creating, shaping, and interacting with music is an active area of research in algorithmic composition and AI music. The field includes a wide range of approaches, including agents, neural networks, machine learning, generative grammars, Markov models, chaos theory, genetic algorithms, cellular automata, and other computational methods. This section presents projects, software demos, research, and creative works that explore these techniques through algorithmic composition, procedural music, and interactive sonic systems.

Publication:

Algorithmic Interactive Music Generation in Videogames: A Modular Design for Adaptive Automatic Music Scoring

A Progressive-Adaptive Music Generator (PAMG): An Approach to Interactive Procedural Music for Videogames”. Proceedings of the 12th ACM SIGPLAN International Workshop on Functional Art, Music, Modelling, and Design (FARM’24)

“Demo: Progressive-Adaptive Music Generator (PAMG) and the Trial Game”. Proceedings of the 12th ACM SIGPLAN International Workshop on Functional Art, Music, Modelling, and Design (FARM’24)