Speech identification under simulated hearing-aid frequency response characteristics in relation to sensitivity, frequency resolution, and temporal resolution

Word identification in noise was measured adaptively under flat and rising frequency response conditions to represent basic alternatives for a hearing-aid characteristic. The speech test results were compared with measures of sensitivity, loudness tolerance, frequency resolution, and temporal resolution in 23 hearing-aid users with mild or moderate sensorineural hearing losses. Subjects also rated the two frequency responses for preference and subjective quality. A paradoxical relationship was found whereby superior speech performance under the flat condition was associated with preference for the rising condition, and vice versa. No combinations of psychoacoustic variables satisfactorily explained either relative performance or preference, although high-frequency sensitivity and upward spread of masking were implicated. Absolute speech performance was related to sensitivity at 2 kHz, age, and sex, but not to frequency resolution once other factors were partialed. Temporal resolution was also a factor, but this was due largely to the influence of extreme values in two subjects. It is concluded that, for moderate degrees of hearing loss, speech identification in noise can be predicted from age, sex, and sensitivity with little advantage from recourse to measurement of frequency or temporal resolution.