Estimation of the Phonetic Speech Quality Using the Information Theoretic Approach

Abstract

The problem of estimation of the speech phonetic quality using a short fragment of the voice signal is formulated. The Kullback–Leibler minimum information discrimination principle is used to propose a novel criterion and develop an algorithm. The dynamic properties of the algorithm are analyzed. The algorithm is employed in the analysis of the functional state of speaker using the voice signal. It is demonstrated that the needed duration of the voice signal is several minutes only.