A New Parameter of Speech Character Based on the Bloomfield´s Model

Bloomfield´s model has rarely been applied to the speech processing since being put forward; this paper proposes a new parameter of speech character based on this model. The simulation results show that this model is better than linear prediction model when predicting speech signals. Specifically, this new model is of the advantages such as less parameters, lower computation complexity and well characterizing speakers. In addition, the higher rate of voiced/unvoiced decision may be obtained during pitch tracking if the speech is preprocessed by the inverse-filter constructed using Bloomfield´s model. All of these show a good prospect of the Bloomfield´s model in speech processing.