A self-report study that gauges perceived and induced emotion with music

• Published in: 2015 Internet Technologies and Applications (ITA) pp. 239 - 244
• Main Authors: Griffiths, Darryl, Cunningham, Stuart, Weinel, Jonathan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-09-2015
• Subjects: affective computing; audio feature extraction; automatic playlist generation; Fires; machine learning; music and emotion; Physiology; Psychology; Rocks; Sleep
• ISBN: 1479980366; 9781479980369
• DOI: 10.1109/ITechA.2015.7317402

This paper discusses a particular study that gauges emotion with respect to music by means of an online self-report survey. This is part of the ongoing construct of an intelligent mobile music player application that will adjudicate one's activity, environmental context, and physiological state. The study was structured so as to acquire emotional information by giving each participant a discrete affect word including `happy', `excited', `angry', `afraid', `miserable', `sad', `tired', and `relaxed'. Each affect word that was selected for a given song was corroborated by a degree of intensity using 5-point scale. The fundamental objective of this study was to measure as to how each song could be described emotionally, and how each song made them feel emotionally. Pearson's Chi-Squared concluded that 95% of the ratings for both `described' and `induced' emotions were statistically significant. The corresponding results were scaled to the proposed circular-based emotional model based on Russell's Circumplex model of emotion, by converting both the emotional ratings to polar coordinates. Further analysis of the data subsequently showed that the affect words that represent music could be given some granularity around the perimeter of the circle by expanding upon particular properties of Russell's circular ordering of affect words. Lastly, this paper concludes with a Section that outlines the future work for this research.