Improved objective measures for characterization of noise suppression algorithms

We present two improved objective measures evolved from earlier work (see Paajanen, E. et al., Proc. IEEE Workshop on Speech Coding, 2000) and one new measure for the characterization of the effect of noise suppression (NS) algorithms. We enhance the formulation of SNRI (signal-to-noise ratio improvement) and NPLR (noise power level reduction) measures by replacing frame powers by their logarithms to improve the robustness of the metrics and to increase the association with subjective perception. A new third measure is the difference between SNRI and NPLR (DSN) that monitors the consistency of an NS solution in attenuating the background noise while causing a minimal effect on the speech level. The earlier metrics are included as characterization tools in recent 3GPP and TIA minimum performance requirement specifications for NS algorithms in mobile terminals. The enhanced and new measures are proposed for a draft ITU-T recommendation, G.VED, "Voice Enhancement Devices". We present the revised formulation of the measures as well as results of a verification study, including a regression analysis between the presented measures and listening test results. The analysis suggests that the objective measures are relevant for the perception of overall acceptability of the effect of an NS algorithm.