A Bio-Inspired Approach to Infer Functional Rules and Aesthetic Goals from Music Genre Styles

Computational Intelligence in Arts is a recent area of research. There is a growing interest in the application of these techniques in fields such as style recognition. Several works have been proposed for the recognition of styles for performers in which the improvisation often plays an important role. However, most of music genres, as classical music, are based on written music often composed according to specific rules. In this complex context, the style is the result of aesthetic goals, i.e. experience and preferences of the main composers in such a genre, and functional rules, i.e., the rules used to compose music according such a style. We propose a new approach for both recognition and automatic composition of music genre styles, which exploits a machine learning recognizer based on one-class support vector machines and neural networks, and a splicing composer. To assess the effectiveness of our system we performed several tests using a large corpus of 4-voice Bach's music. About the recognition, we show that our classifier is able to achieve an accuracy of 96.2%. With regard to the composition, we measured the capability of our system to capture both aesthetic goals, by collecting subjective perceptions from domain experts, and functional rules by computing the average percentage of classical forbidden exceptions that occur in the compositions produced.