Symbolic regression as a computer-aided music tool for analysis and composition

The paper presents a new OpenMusic library that implements a Genetic Programming method of Symbolic Regression on sets of input data-points and seeks for a Common Lisp function (S-expression) that can be used either to create mathematical models that could potentially help to understand the mathe- matical behavior of the input data or to generate parameters in computer-aided composition. Stressing that a number of issues must still be addressed to im- prove the proposed library, the paper presents some of the strategies to do this and to make Symbolic Regression a practical tool in computer-assisted music composition and analysis.