Quantitative analysis of F0 contours of emotional speech of Mandarin

The F0 characteristics of Mandarin speech in four basic emotions (anger, fear, joy, and sadness) as well as in neutral reading are compared quantitatively. Two approaches are employed: analysis of surface features from time-normalized F0 contours, and analysis-by-synthesis of time-intact F0 contours based on the command-response model, which turns out to be also applicable to emotional speech. For surface F0 features, the height and range of F0, the local tonal variation, and the sentential F0 declination are all investigated. In model-based analysis, the parameters of both phrase and tone commands are compared systematically. The study shows that those surface F0 phenomena can be explained better by the model-based approach, which can later be used in F0 generation for emotional speech synthesis.