Perceptual Study for Emotional Speech of Mandarin Chinese

This paper conducts a set of perceptual experiments together with an F0 acoustic analysis on emotional speech of Mandarin Chinese. Sixty utterances spoken by two actors in five basic emotions (i.e. happiness, fear, anger, sadness, and boredom) as well as in the neutral style served as stimuli. The perceptual experiments showed the following results: (1) For Mandarin Chinese, the rates of identification for five basic emotions differ significantly, ranking in Sadness > Happiness > Anger > Fear > Boredom. (2) Perceptual confusion occurs mainly between fear and sadness, and between boredom and anger. There is a perceptual boundary within each pair. The similarity coefficient is 0.16 for fear and sadness, and 0.22 for boredom and anger. (3) The accuracy in the perception of emotions varies with the linguistic background of the listener; non-native subjects show a degraded perception. Moreover, acoustic analysis implies that F0 feature is not the only cue for the perception of emotions in Mandarin Chinese.