Distinctive feature analysis of binaural versus monaural consonant perception as a function of signal to noise ratio

Electret microphones inserted in the ear canals were used to binaurally record noncorrelated noise from two speakers equidistant at ±45° azimuth in an anechoic chamber. Twenty-one CV's (with “a) were similarly recorded at 0° azimuth. Various signal to noise ratios were obtained by mixing each binaural recording. These recordings maintained the effects of head shadow, pinna, etc. At each signal to noise ratio subjects wrote their response to the 30 presentations of each CV binaurally and monaurally. Confusion matrices were constructed. Data were analyzed in terms of consonant confusions and feature intelligibility. Multivariate analysis indicated differences in the contribution of the various features when binaural and monaural data were considered and also revealed changes in this pattern as a function of signal to noise ratio.