Pollack, I (1948). Effects of high-pass and low-pass filtering on the intelligibility of speech in noise. J. Acoust. Soc. Am. 20, 259-266

The effect on intelligibility of eliminating either the high frequency speech sounds or the low frequency speech sounds was determined by standardized articulation testing procedures under the special circumstance of a background of white masking noise. In general, it was found that intelligibility increased as the frequency range and the intensity level of the speech signal were increased. Contours are presented which describe the interrelations among (1) the frequency range of the speech signal, (2) the intensity level of the speech signal, and (3) articulation efficiency. For each of the experimental conditions, the articulation index - a measure developed by the Bell Telephone Laboratories - was computed. It was found that the scatter of the computed indices, when plotted against the scores actually obtained, was of the same order of magnitude as the intrinsic variability of typical articulation scores. This result indicates that the computational procedures are an acceptable substitute for time-consuming articulation testing when an approximate assessment of the relative efficiencies of a series of communication systems is required and when a background noise is encountered like the white noise of this study. It was found that the relative contribution of the various speech frequencies was not constant, but rather was a function of the intensity of the speech signal relative to the constant white noise masking signal. More specifically, as the intensity level of the speech signal was increased, the relative contribution to intelligibility of the higher speech frequencies increased.