Music Recommendation Based on Color

Abstract

This study presents an alternative way in choosing songs based on a selection of colors through Color-to-Music application. There were three stages of this study: The first stage was the preparation music library of the association between color and emotion; and the association between music and emotion. Library data used for the Hue, Saturation, and Value (HSV) color model creation were: Hue to represent musical instruments, Saturation referred to tempo, and Value was key (pitch). Second stage was to create two types of graphical user interface (GUI) for color selection. The last stage was to collect data from 120 participants participating in the trials. This study focused on two questions. First was the accuracy rate of a recommended song that matched selected colors. Second was to find the most suitable GUI that provides the highest accuracy rate to the recommendation of songs. The tests were divided into two groups: test A and test B. As for test A, participants started a trial by choosing color from the application; while test B would start by selecting an initial emotion, and then choosing colors to match the chosen emotions. The results showed that the overall accuracy rate of test A is higher than test B and the slider GUI has the highest accuracy rate.