Building an Initial Fitness Function Based on an Identified Melodic Feature Set for Classical and Non-Classical Melody Classification

Algorithmic Composition for music is a progressive field of study. The success of an automated music- generating algorithm, however, depends heavily on the fitness function that is used to score the generated music. Artificial intelligence in this context of music scoring can use a fitness function that is generally based on music features that a given algorithm is programmed to measure. This study explores the features that are important for melody generation by investigating those that can separate classical from non-classical music in the context of melody. The jSymbolic tool was used to collect 160 standard features from 400 music files. Using C4.5 algorithm to select significant features used in a classical vs. non-classical melody classification challenge, and then performing a comparison with a suggested feature set by running Naïve-Bayes and SVM classifiers, the study was able to determine a candidate set of melodic features that can be used for building an initial fitness function that separates classical from non-classical melodies with high accuracy as revealed by SVM ten-fold cross validation.