Abstract

This work investigates a structurally grounded approach to algorithmic music composition using a stochastic Context Free Grammar (CFG) evolved through a genetic algorithm. While existing systems such as live-coding environments, DAWs (Digital Audio Workstations), and statistical models generate musical material, they typically lack explicit hierarchical organization, despite evidence that listeners perceive music sequentially and structurally. In this framework, derivation trees generated from an CFG serve as compositional candidates, which are evolved using subtree-based crossover and mutation. Fitness evaluation combines prior melodic heuristics with tension-based insights from music cognition research, introducing musically informed structure into the evolutionary process. A genetic system was developed that can generate chord progressions and be able to distinguish from those that have more or less presence of perceived tension and relaxation, according to human studies, which follows a closer reflection of how listeners understand musical form.

Publication Date

4-24-2026

Document Type

Thesis

Student Type

Graduate

Degree Name

Computer Science (MS)

Department, Program, or Center

Computer Science, Department of

College

Golisano College of Computing and Information Sciences

Advisor

Thomas J. Borrelli

Advisor/Committee Member

Matthew Fluet

Advisor/Committee Member

Joe Geigel

Campus

RIT – Main Campus

Download

Share

COinS