Effect and Sound Localization with Dichotic-Listening Digital Hearing Aids

Sensorineural hearing loss is commonly accompanied by reduced frequency selectivity, which causes the larger adverse effect of masking. To cope with this issue, some studies have addressed the effects of dichotic listening to reduce the masking between contiguous frequency bands. This paper investigates sound image localization perceived in both dichotic and diotic listening conditions, as well as speech intelligibility. The first experiment carried out a speech intelligibility test using vowel-consonant-vowel nonsense syllables presented to four subjects with mild to severe sensorineural hearing loss. Speech stimuli were divided into two frequency bands for dichotic listening to reduce masking to a consonant by the preceding vowel. In consideration of formant frequencies of Japanese vowels, the frequency boundary was set to 0.8 kHz (between F1 and F2) and 1.6 kHz (middle of F2). Results showed that greatest improvement in intelligibility was observed when the dividing frequency was 0.8 kHz. The second experiment conducted sound image localization using daily conversation speech samples for six young adults with normal hearing and three elderly persons with hearing impairment. Head-related transfer functions (five directions, non-individualized) were convolved to realize directivity of the speech samples. Results show that localization errors varied with the dividing conditions and whether hearing was impaired or not.