Hybrid method of speech signals spectral analysis based on the autoregressive model and Schuster periodogram

Abstract

The task of measuring the power spectral density of a speech signal in the regime of a sliding observation window is considered. A parametric approach to solving this task using an autoregressive data model is studied. The problem of optimizing the order of an autoregressive model under conditions of small samples is studied. The problem is proposed to be solved using a hybrid method of spectral analysis based on sequential enumeration of a fi nite number of alternatives. The optimization criterion is formulated in terms of the inverse problem: from a speech signal to a voice source. It uses the scaleinvariant measure of the spectral distance as the objective function, and the Schuster periodogram as the reference sample. The effectiveness of the hybrid method has been experimentally evaluated on the basis of author's software. It is shown that with the duration of the observation window of no more than 10 ms, the use of the hybrid method increases the accuracy of spectral analysis by more than 30 % compared to the well-known Berg method, the order of which is established according to the Akaike information criterion.