In this study, a method of tracking the formant trajectory of Tracheoesophageal (TE) speech using the Hidden Dynamic Model (HDM) with dynamic target orientation was put forward based on the characteristics of TE speech. At first, a PIF-LPC algorithm was used to extract the formant parameters of TE speech. The parameters as a weighted dynamic component were then introduced in HDM; afterwards, HDM was solved using the particle filtering algorithm based on the prior distribution of this parameter to track formant trajectory. It was shown in simulation tests that the method was of high precision and good robustness owing to the orientation effect of dynamic targets. With this method, not only the interferences of spurious peaks and merged peaks in TE speech were overcome, but also continuous TE speech signals were tracked.
展开▼
