Applying Affect Estimation to 3D Music Visualization

Abstract

This project aimed to develop a 3D music visualizer using affect estimation and real-time audio features. Methods used during the project ranged from research through design, focus groups, prototyping, and an experimental survey. The final iteration of the visualizer could generate 3D scenes based on the music and extracted affect values, which participants found to be visually pleasing and fitting to the music. However, the visualizer struggled to accurately communicate the affect values visually, which is attributed to the lacking aesthetic profile and the subsequent implementation. Despite this, the project successfully manifested an affect-based visualizer that offers potential for further development. Future research is needed to investigate the use of abstract means for communicating emotions, and standardization of measures and language is required for this multidisciplinary domain.