Pitch extraction using dyadic wavelet transform and modified higher order moment

Abstract

Pitch detection is the process of determining the period of the vocal cords closure or in another word the time duration of one glottal closed, open and returning phase. Dyadic wavelets transform (DyWT) and modified higher order moment, which is based on the autocorrelation function, are two pitch detection methods. DyWT is an accurate pitch detection method, however it has less accuracy compared to modified higher order moment. On the other hand modified higher order moment has high computational complexity and is time consuming. In this paper, we propose a pitch detection method based on DyWT which has use modified higher order moment. Modified higher order moment is applied only in some specific parts to improve the accuracy and keep the computational complexity low. Finally, our simulations and evaluations on some speech utterances indicate that the new pitch detection method improves previous DyWT method.