Statistical Conversion Algorithms of Pitch Contours Based on Prosodic Phrases

Pitch contour of a speech utterance plays an important role in expressing speaker's individuality and meaning of the utterance. In performing speech conversion from a source speaker to a target speaker, it is important that the pitch contour of the source speaker’s utterance be converted into that of the target speaker. This paper investigates statistical algorithms of pitch contour conversion for Korean language. The algorithms are based on Gaussian normalization, and its combination with a declination-line modeling of pitch contour. Pitch contour conversion are investigated at two levels of prosodic phrases: intonation phrase and accentual phrase. Experimental results show that the algorithm of Gaussian normalization within accentual phrases is significantly more accurate than the algorithms for intonational phrases in pitch contour conversion.