Visimu: a game for music color label collection

Abstract

Based on previous studies of the associations between color and music, we introduce a scalable way of using colors to label songs and a visualization of music archives that facilitates music exploration. We present Visimu, an online game that attracted users to generate 926 color labels for 102 songs, with over 75% of the songs having color labels reaching high consensus in the Lab color space. We implemented a music archive visualization using the color labels generated by Visimu, and conducted an experiment to show that labeling music by color is more effective than text tags when the user is looking for songs of a particular mood or use scenario. Our results showed that Visimu is effective to produce meaningful color labels for music mood classification, and such approach enables a wide range of applications for music visualization and discovery.