A method for predicting speech sound recognition performance by the hearing impaired

We have reported earlier [J. Acoust. Soc. Am. 74, 1606–1612 (1983); J. Acoust. Soc. Am. Suppl. 1 75, S84 (1984)] the details of an articulation index (AI) scheme that was found satisfactory for individual normal-hearing listeners. We have also reported on the difficulties associated with applying the AI procedure to individuals with hearing impairment [J. Acoust. Soc. Am. 75, 1253–1258 (1984)], as well as the results of our earlier attempts to introduce modification in the procedure to account for hearing loss [J. Acoust. Soc. Am. Suppl. 1 74, S104 (1983)]. We now present the theoretical framework of an approach that we found adequate for a number of hearing impaired subjects with different audiometric configurations. The procedure is based on relating various psychoacoustical variables to each other and to the speech processing capacity of the auditory channel. The method was tested and found satisfactory under various conditions of noise distortion, speech filtering, and low presentation levels. [Work supported by NINCDS.]