Music compositional intelligence with an affective flavor

Abstract

The consideration of human feelings in automated music generation by intelligent music systems, albeit a compelling theme, has received very little attention. This work aims to computationally specify a system's music compositional intelligence that tightly couples with the listener's affective perceptions. First, the system induces a model that describes the relationship between feelings and musical structures. The model is learned by applying the inductive logic programming paradigm of FOIL coupled with the Diverse Density weighting metric over a dataset that was constructed using musical score fragments that were handlabeled by the listener according to a semantic differential scale that uses bipolar affective descriptor pairs. A genetic algorithm, whose fitness function is based on the acquired model and follows basic music theory, is then used to generate variants of the original musical structures. Lastly, the system creates chordal and non-chordal tones out of the GA-obtained variants. Empirical results show that the system is 80.6% accurate at the average in classifying the affective labels of the musical structures and that it is able to automatically generate musical pieces that stimulate four kinds of affective impressions, namely, favorableunfavorable, bright-dark, happy-sad, and heartrending-not heartrending.

Exploring Melodic Motif to Support an Affect-Based Music Compositional Intelligence

“…1, wherein CH is a candidate chromosome) takes into account what the induced relations indicate as fit to the user's affect perceptions as well as fit to certain principles in music theory that we specified (discussed in detail in [5]). This makes it possible to generate chord progressions that fit the music theory and cause the perceived affect.…”

“…This motivation to translate human subjective perceptions to product design parameters originated from Kansei (this is a Japanese word that literally means human feelings; researchers in the west later on provided the equivalent term Affective) Engineering [3], [4], which is the study of human-product interactions by quantifying and measuring human affect (i.e., mood or emotion) and its correlation to certain product properties, and to use this correlation knowledge to design products that are more satisfying to consumers. The CAUI modifies and creates new music score features, hence constructive, after these score features are classified into some affective perceptual categories [5]. It is adaptive since it recognizes and adjusts its compositional knowledge based on user affective perceptions.…”

“…Since its conception, the design of the CAUI has been continuously developed on several fronts [2], [6], [5], [7], namely, target outcome (e.g., predictive models of user affect perceptions, musical arrangements, and original music compositions), scope of music theory (e.g., partial and/or overall music structures, chord functions, and 8-bar chord progressions), machine learning algorithms, (e.g., refinement theory, genetic algorithm, diverse density with a weighting metric applied over multiple-part data, and symbiotic evolution [8]), and affect labeling instruments (e.g., semantic differentials and EEG-based emotion spectrum analyzer [9]). A persisting and major shortcoming of the CAUI, however, is creating adequately well-structured tunes from the viewpoint of music theory, particularly in coming up with compositions that are more fluent, cohesive and melodic.…”

“…Lastly, melodies are added by modifying the best chordal tones that the GA found in order to create non-harmonic tones thereby creating non-monotonic musical pieces. We refer the reader to our earlier paper, [5], that discusses in greater length the technical aspects of the framework. We note however two differences in the framework, namely, that we only used here FOIL and without the Diverse Density algorithm and improved the genetic algorithm component by incorporating the minimal generation gap technique.…”

The Shopping Experience of Tomorrow: Human-Centered and Resource-Adaptive

“…Instead of attracting the listener's attention, the soundscape should have a calming and mood influencing effect (see also [4,35]). …”