Voice quality perception from emotional voices: Comparison of emotional utterances in English and Japanese

Voice quality refers to the auditory impression (e.g., hoarse and nasal) the listener experiences from a piece of speech, and it provides information about the speaker such as his/her physical characteristics, age, and emotional state. The present study aimed to develop a scale in English language to measure how we perceive qualities of emotional voices expressing happy, angry, sad and fear, based on an existing Japanese scale [1]. Further, it examined whether the voice qualities related to those emotions differ between Japanese listeners and American listeners. In Study 1, English adjectives used to describe voices emoting happiness, anger, sadness, and fear were collected from native English speakers, and then ten words were selected for the scale. In Study 2, participants from the US and Japan listened to voices expressing the four emotions and evaluated qualities using the scales in their native language. A factor analysis and a canonical discriminant analysis on the data showed that the results from the two cultures were very similar despite the difference in adjectives contained in the scales. The ten words in each scale were grouped into three factors and all factors matched closely for the two cultures. The canonical discriminant analysis revealed three discriminant functions available to distinguish different emotions based on the quality ratings. These dimensions appear to indicate variations in valence and activation of the expressed emotion, and tensions of the speakers’ vocal folds. By using these dimensions, the voice quality ratings can distinguish emotions of the voice fairly well, demonstrating that the perception of voice quality is an important aspect of recognizing emotion from voices.