A Survey on Emotional Models and Music Features in Music Emotion Recognition Systems

PhD Qualifying Examination


Title: "A Survey on Emotional Models and Music Features in Music Emotion 
Recognition Systems"

by

Mr. Bing Yen CHANG


Abstract:

Music emotion and informatics research has been growing alongside the 
popularity of music streaming software and services. Of particular 
interest are Music Emotion Recognition (MER) systems, which can 
automatically label the emotional content of music tracks via feature 
extraction and subsequent classification. This survey summarizes the 
emotional models and features which comprise the framework of MER systems. 
Both the dimensional and categorical emotional models are discussed and 
compared, with the Valence-Arousal model most widely used in the 
literature. Common features, their extraction methods, as well as their 
correlations with music emotion are also reviewed. The features can be 
categorized into three domains – Musical, Temporal and Spectral, and 
Machine Learning – forming a tradeoff between interpretability and 
generalization power. Lastly, the nuanced coverage of the music emotion 
space by these features is discussed, with possibilities for further 
research.


Date:			Tuesday, 17 December 2019

Time:                  	9:15am - 11:15am

Venue:                  Room 2132C
                         Lift 19

Committee Members:	Prof. Andrew Horner (Supervisor)
 			Dr. Raymond Wong (Chairperson)
 			Dr. David Rossiter
 			Dr. Pedro Sander


**** ALL are Welcome ****