8th Annual Conference of the International Speech Communication Association

Antwerp, Belgium
August 27-31, 2007

Analysis of Emotional Speech Prosody in Terms of Part of Speech Tags

Murtaza Bulut, Sungbok Lee, Shrikanth S. Narayanan

University of Southern California, USA

Representation of emotions in terms of acoustic features of well defined lexical elements is desired for development of emotional speech processing systems. For that purpose, in this paper, the interaction between emotions and part of speech (POS) tags is investigated. Utterances from 3 speakers in angry, happy, sad, and neutral emotions are used to statistically analyze the effects of emotion, POS tag type, position of the tag, and speaker factors on tag duration, energy, and F0 variables. It is found that the main effects of emotion, tag type, and position are significant. Results also show that the effect of emotion is significantly dependent on position, but not on POS tag type. The effect of position is noticeable. POS tags located in the first half of sentences have shorter durations, higher energy, and higher F0 values.

Full Paper

Bibliographic reference.  Bulut, Murtaza / Lee, Sungbok / Narayanan, Shrikanth S. (2007): "Analysis of emotional speech prosody in terms of part of speech tags", In INTERSPEECH-2007, 626-629.