Improving parametric trajectory modeling by integration of pitch and tone information

This paper presents the application of pitch/tone information to improve Parametric Trajectory Modeling (PTM). To simulate the trajectory of pitch in a segment in PTM recognizer, we in fact get its corresponding tone information. From another point of view, tone is a segmental feature and PTM has the excellent framework to incorporate it. So we here introduce the “soft” and “hard” integration methods to combine them with the base modeling. In the experiment of Mandarin digit classification, they get 22.87% and 33.54% error reduction respectively, and integration of both them together obtains 38.72% error reduction.