Paper to be presented at XVIIIth ANNUAL CONFERENCE OF INDIAN ASSOCIATION OF BIOMEDICAL SCIENTISTS

Sensorineural hearing loss causes reduction in frequency resolving capacity of the ear due to spread of masking of frequency components by adjacent frequency components. We have carried out experimental evaluation of splitting speech into two complementary signals on the basis of frequency and presenting it binaurally for increasing the speech intelligibility. In this scheme, input speech signal is filtered into two signals by using a bank of critical band filters where odd numbered critical bands are presented to one ear and even numbered ones to the other. Thus, the effect of spectral masking in the cochlea on speech information reaching the auditory cortex is reduced. The process of splitting speech was done off-line, and listening tests were performed using twelve English consonants in vowel-consonant-vowel context with /a/ as in `father’, on four normal hearing subjects with simulated sensorineural hearing loss. Simulation of sensorineural impairment in subjects with normal hearing was done by addition of white masking noise to the speech signal at different SNRs. The test results were analyzed by comparing the recognition scores for various unprocessed speech to those with processed speech. Under adverse masking noise condition, the recognition score is found to increase by 10 percent. The average response time over four subjects is also found improved by this scheme.