Auditory filtering and the discrimination of spectral shapes by normal and hearing-impaired subjects

A review of the literature suggests that many hearing-impaired patients suffer from sensory deficits in addition to the reduced audibility of speech signals. Poor frequency resolution, or abnormal spread of masking, is a consistently identified deficit in sensorineural hearing loss. Frequency resolution was measured in individual subjects using the input filter pattern paradigm, and the minimum detectable amplitude of a second-formant spectral peak in a spectral-shape discrimination task was also determined for each subject. The two tasks were designed to test the identical frequency regions in each subject. A nearly perfect correlation was found between the degree of frequency resolution as measured by the input filter pattern and performance on the spectral-shape discrimination task. These results suggest that measures of frequency selectivity may offer predictive value as to the degree of impairment that individual hearing-impaired patients may have in perceiving the spectral characteristics of speech, and also lead to suggestions for signal processing strategies to aid these patients.