Predicting Human Reactions to Music on the Basis of Similarity Structure and Information Theoretic Measures of the Sound Signal

Memory, repetition, and anticipatory structure are considered important characteristics of musical style. Both composers and listeners often refer to these parameters in describing the music. In this work we conducted audio analyses in an attempt to determine correlations between audio features and human responses of Familiarity and Emotional Force. The analyses were performed on two recordings of an extended contemporary musical composition by one of the authors. Signal properties were analyzed using statistical analyses of signal similarity structure over time and information theoretic measures of signal predictability. The analyses show strong evidence that signal properties and human reactions are related. Application to style analysis are discussed.