Evaluation of Speech Processing Schemes using Binaural Dichotic Presentation to Reduce the Effect of Masking in Hearing-Impaired Listeners

Binaural dichotic presentation schemes help in increasing the intelligibility of speech for persons with moderate bilateral sensorineural hearing impairment. Spectral splitting, with comb filters has helped in improving the perception of place feature. Temporal splitting, using trapezoidal fading functions, provided improvement in the perception of duration feature. Splitting of speech using time-varying comb filters, in which the bands of the comb filter are cyclically swept to have spectral and temporal splitting simultaneously, provided improvement in features of place and duration, without affecting the other features. An overall evaluation has been carried out, by conducting listening tests on seven normal hearing subjects with simulated hearing loss, on the three processing schemes: (i) spectral splitting with comb filters based on auditory critical bands designed for minimum spectral distortion and perceptual balance, (ii) temporal splitting with interaural switching of 20, 40, and 80 ms with trapezoidal fading functions of 70 % duty cycle and 3 ms transition, and (iii) combined spectral and temporal splitting using time-varying comb filters with sweep cycles of 20, 40, 80, 120 and 160 ms, each with 4, 8, and 16 shiftings. Test material consisted of phonetically balanced monosyllabic words in listener's native language. Hearing loss was simulated by adding broadband noise at different levels to the speech stimuli at constant shorttime SNR of 10 ms. At recognition score of 60 %, spectral splitting provided an improvement of 5 dB in SNR. Combined splitting with sweep cycle of 80 ms or greater with 8 and 16 shiftings provided almost the same improvement.