School of Music & Music Education

Home > About us > Research

Emotion in Music

The ability of music to express and produce emotions is one of its most fascinating properties. Some of the research projects which aim to better understand how music expresses and evokes emotion, and what emotions it expresses are:

Measuring Emotion Continuously on a Two Dimensional EmotionSpace (2DES)
Emotion Face
Musical Features
The People

Measuring Emotion Continuously on a Two Dimensional EmotionSpace (2DES)

Emotion can be described as being a multidimensional construct. Several researchers agree that two salient , bipolar dimensions of emotion are the valence (happiness-sadness) and arousal (aroused-sleepy). Software for measuring emotional responses to music (and other stimuli) continuously has been developed and tested [Software is currently being updated and will be available soon. An old, free version for Power Mac (mid 1990s macintosh computers with CD drives) can be downloaded here)]

Selected References:

Schubert, E. (2003). Update of Hevner's Adjective Checklist.   Perceptual and Motor Skills. 96, 1117-1122.  

Schubert, E. (1999). Measuring Emotion Continuously: Validity and Reliability of the Two Dimensional Emotion Space.   Australian Journal of Psychology, 51, 154-165.

Schubert, E.   (1996). Measuring Temporal Emotional Response to Music Using the Two Dimensional Emotion Space.   Proceedings of the 4th International Conference for Music Perception and Cognition , Montreal, Canada (11-15 August), 263-268.

EmotionFace [top]

Once emotional responses to music are collected by the 2DES, how can they be displayed? This projects explores the possibility of using simple sketches of changing facial expressions to represent the emotion expressed by a piece of music.

1. Click here [2MB] to see the EmotionFace as it responds to an excerpt of the slow movement from Concierto de Aranjuez by Joaquin Rodrigo. Sadness in the music is reflected by the up-side-down parabolic mouth shape.

2. Click here [3.7MB] to see the EmotionFace as it responds to an excerpt of 'Morning' from Peer Gynt by Edvard Grieg. Notice the gradual widening of the eyes (increasing arousal) as the piece gets louder.

3. Click here [2MB] to see the EmotionFace as it responds to an excerpt of Slavonic Dance No. 1, Op. 46 by Antonin Dvorak. The sudden burst at the beginning produces a startle-like effect in the 'listener', who after a few seconds realises that this is a joyous piece.

For more information about how this face is constructed, read the paper to be presented at the International Conference on Auditory Display

Musical Features [top]

From the above projects, we have evidence that there is significant agreement in the emotion that a piece of music can express within a given culture (the culture represented by those whose emotional perceptions were recorded). It should therefore be possible to determine what musical features vary to produce these emotional responses.

Using continuous response methodology and time series analysis techniques, it has been possible to produce regression equations which model emotional responses as a function of musical features alone [reported in Schubert, E. (2004).   Modeling perceived emotion with continuous musical features.   Music Perception, 21(4), 561-585 - Abstract follows]:

Abstract

The relationship between musical features and perceived emotion was investigated using continuous response methodology and time-series analysis.   Sixty-seven participants responded to four pieces of Romantic music expressing different emotions. Responses were sampled once per second on a two dimensional emotion space (happy-sad valence and aroused-sleepy).   Musical feature variables of loudness, tempo, melodic contour, texture, and spectral centroid (related to perceived timbral sharpness) were coded.   Musical feature variables were differenced and used as predictors in two univariate linear regression models of valence and arousal for each of the four pieces.   Further adjustments were made to the models to correct for serial correlation.   The models explained from 33% to 73% of variation in univariate perceived emotion.   Changes in loudness and tempo were associated positively with changes in arousal, but loudness was dominant.   Melodic contour varied positively with valence, though this finding was not conclusive.   Texture and spectral centroid did not produce consistent predictions.   The methodology facilitates a more ecologically valid investigation of emotion in music and, importantly in the present study, enabled the approximate identification of the lag between musical features and perceived emotion.   Responses were made one to three seconds after a change in the causal musical event, with sudden changes in loudness producing response lags from zero (nearly instantaneous) to one second.   Other findings, interactions and ramifications of the methodology are also discussed.

Selected References

Schubert, E. (2004). Modeling perceived emotion with continuous musical features. Music Perception, 21(4), 561-585 .  

Schubert , E. (2002). Correlation Analysis of Continuous Emotional Response: Correcting for the effects of serial correlation. Musicae Scientiae, Special Issue 2001-2002, 213-236 .  

Schubert, E. (2001). Continuous Measurement of Self-Report Emotional Response to Music.   In P. Juslin and J. Sloboda (Eds.), Music and Emotion: Theory and Research. (pp. 393-414) Oxford University Press.  

Schubert , E. & Dunsmuir, W. (1999). Regression modelling continuous data in music psychology.   In Suk Won Yi (Ed.), Music, Mind, and Science (pp. 298-352). Seoul National University Press.

Schubert, E. (2004). Research in expressing continuous emotional response to music as a function of its acoustic parameters:   Current and future directions. Invited paper for the   18th International Congress on Acoustics (ICA 2004). April 4-9, 2004, Kyoto International Conference Hall, Kyoto, Japan.

The people [top]

The emotion in music projects are led by Dr. Emery Schubert. The projects have involved either generous assistance from or collaboration with Prof. William Dunsmuir, Assoc. Prof. Gary McPherson, Prof. Joe Wolfe (Music Acoustics Group, UNSW), Dr. Dorottya Fabian, Dr. Kate Stevens (MARCS and AMPS).

For more information, please contact Emery Schubert (E.Schubert@unsw.edu.au) .