Comparing emotions using acoustics and human perceptual dimensions

ABSTRACT

Understanding the difference between emotions based on acoustic features is important for computer recognition and classification of emotions. We conducted a study of human perception of six emotions based on three perceptual dimensions and compared the human classification with machine classification based on many acoustic parameters. Results show that the six emotions cluster differently according to acoustic features and to perceptual dimensions. Acoustic features fail to characterize the perceptual dimension of valence. More research is needed to find acoustic features that have a close relation to human perception.

REFERENCES

• 1: Borg, I., and Groenen, P. Modern Multidimensional Scaling: Theory and Application. Springer, New York, USA, 2005.
• 2: Dai, K., Fell, J.H., and MacAuslan, J. Recognizing Emotion in Speech Using Neural Networks. In Proc. IASTED 2008 on Assistive Technologies, ACTA Press (2008), 31--36.
• 3: Darwin, C. The Expression of Emotion in Man and Animals. John Murray, London, UK, 1872.
• 4: Kwon, O., Chan, K., Hao, J., Lee, T. Emotion Recognition by Speech Signals. In Proc. EUROSPEECH 2003, ISCA (2003), 125--128.
• 5: Laukka, P. Vocal Expression of Emotion: Discrete-Emotions and Dimensional Accounts. PhD thesis, 2004, Uppsala University (2004).
• 6: Lee, C. M., Yildirim. S., Bulut, M., Kazemzadeh, A. Busso, C., Deng, Z., Lee, S., and Narayanan, S. Emotion Recognition Based on Phoneme Classes. In Proc. ICSLP 2004, ISCA (2004), 889--892.
• 7: Lewis M., and Hayiland J.M. Handbook of Emotions. Guilford Press, New York, USA, 1993.
• 8: Linguistic Data Consortium (LDC2002S28), Emotional Prosody Speech, www.ldc.upenn.edu/Catalog/CatalogEntry.jsp?catalogId=LDC2002S28, University of Pennsylvania.
• 9: Liu, S., Landmark Detection of Distinctive Feature-Based Speech Recognition. Journal of the Acoustical Society of America, 100(5), 1996, 3417--3430.
• 10: Donn Morrison , Ruili Wang , Liyanage C. De Silva, Ensemble methods for spoken emotion recognition in call-centres, Speech Communication, v.49 n.2, p.98-112, February, 2007 [doi>10.1016/j.specom.2006.11.004]

INDEX TERMS

Primary Classification:
  H. Information Systems
  H.1 MODELS AND PRINCIPLES
      H.1.2 User/Machine Systems

General Terms:
Human Factors

Keywords:
acoustics, emotion classification, emotional speech, human perception of emotion