Self-Organizing Map for Automated Quality Assessment in Algorithmic Music Computation

This work explores the application of Self-Organizing Maps (SOMs) to the domain of algorithmic music creation, performing the role of the most critical element in the creation process, namely, automated quality assessment of the music under composition. Leveraging principles from Rasa theory, which emphasizes the emotional and aesthetic dimensions of universal human experience, our study introduces a novel approach to compose music using specific aspects of Evolutionary Algorithms. It aims to generate music aligned with each of four important Rasas, which will elicit the specific human emotions associated with that Rasa. This innovative approach aims to bridge the gap between computational creativity and human perception, providing a valuable understanding for musicians, composers, and researchers in the field of music. The prime challenge is to synthesize a Fitness Function (FF) that can accurately detect the “degree of adherence” of the composition to the targeted Rasa. Computational results presented here demonstrate the surprising effectiveness of our approach of taking Self-Organizing Map as FF in capturing the intricacies of Rasa classification, over other established and popular approaches.