Improved Objective Measurements for Speech Quality Testing

摘要

This report presents the results of a large study of new objective measures for speech quality. The study defined a figure-of-merit for a particular objective quality measure using correlation analysis between a data base of subjective speech quality measures and a data base of objective speech quality measures. Both data bases were derived from approximately twenty hours of coded and distorted speech. The subjective test used was the Diagnostic Acceptability Measure (DAM) developed at the Dynastat Corp. The study consentrated on three classes of objective measures: those based on a model of the ear, those based on parametric subjective quality estimate, and those based on preclassified distortion sets. All three classes of objective measures performed well, but the best results were obtained for the class of parametric measures.