An Analysis of Objective Measures for User Acceptance of Voice Communications Systems

摘要

The report presents the results of a large study of the statistical correlation 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 eighteen hours of coded and distorted speech. The subjective test used was the Diagnostic Acceptability Measure (DAM) test. The objective measures included spectral distance measures, frequency variant spectral distance measures, a signal-to-noise measurements, area ratio distance measures, log area ratio distance measures, PARCOR distance measures, log PARCOR distance measures, feedback coefficient distance measures, log feedback coefficient distance measures, residual energy ratio distance measures, and composite measures. The analysis procedures included linear regression analysis, multiple linear regression analysis, and nonlinear regression analysis. In all, approximately 1,500 variations of these objective measures were studied. The figure-of-merit used for measuring the performance of an objective measure was the estimated correlation coefficient between the objective measure and the subjective data base. Parametrically different distance measures were compared using nonparametric pairwise rank statistics. The results of this study give quantitative predictions of the performance of many objective speech quality measures for predicting subjective user acceptance.