15th Annual Conference of the International Speech Communication Association

September 14-18, 2014

Mapping Emotions into Acoustic Space: The Role of Voice Quality

Ting Wang (1), Hongwei Ding (1), Jianjing Kuang (2), Qiuwu Ma (1)

(1) Tongji University, China
(2) University of Pennsylvania, USA

This study aims to evaluate the importance of phonation cues in the acoustic realization of four vocal emotions (happiness, fear, anger, and sadness) in Mandarin Chinese. Our perception experiment confirmed that native listeners could differentiate these emotions. To explain how listeners accomplished the task in the perception experiment, we investigated the acoustic cues used by speakers. An acoustic analysis revealed that prosodic measures such as F0, intensity, and duration failed to separate certain emotions in the acoustic space, and thus could not explain how native listeners can perceive the different emotions. However, by incorporating phonation-related cues in the multi-dimensional scaling acoustic space, different emotions were separated clearly. Principal components analysis further revealed the specific contribution of each acoustic measure. These results also allow some preliminary conjectures on how these acoustic components might represent the underlying emotion dimensions, namely, arousal, valence, and control.

Full Paper

Bibliographic reference.  Wang, Ting / Ding, Hongwei / Kuang, Jianjing / Ma, Qiuwu (2014): "Mapping emotions into acoustic space: the role of voice quality", In INTERSPEECH-2014, 1978-1982.