Word Discrimination in Noise

Abstract

This study reports significant differences in listeners' performance when different speakers administer a test of speech intelligibility in noise in a manner that purportedly would yield equivalent test results. Six speakers, three female and three male, each recorded list 4 of the CID W-22 discrimination test. The recordings were mixed with speech noise at speech-to-noise ratios of +5, +1, and -3 dB. Three groups of 14 normal-hearing subjects were used. Each group heard the six speakers at one of the three speech-to-noise conditions. Equivalence between speakers was considered to have been achieved when means were not different and correlations were high. Results suggest that the assumption of interspeaker equivalence may not be valid when speech discrimination is measured in a background of noise.