[A study on digitally processed sounds designed to improve speech sound perception]

This study was performed to improve the speech sound perception of patients with sensorineural deafness by using digitally processed sounds. The fourteen CV sounds (/pa, ta, ka, ba, da, ga, ha, sa, za, ma, na, ra, ja, wa/) were selected in this study, and the consonant burst and/or voice onset time (VOT) of these sounds were doubled and/or amplified by digital processing and stored in DAT. These processed sounds and the original unprocessed sound were presented to patients and with moderate sensorineural deafness. The following results were obtained from patients who had made a mistake in discriminating the original sounds. 1. The correct answer rate for /ta/ and /sa/ was improved by amplification of the consonant burst or VOT, and for /ka/, amplification and/or repetition improved the corredt answer rate. Amplification of the consonant burst, or of the consonant burst with VOT, was especially effective for unvoiced explosive sounds (/pa, ta, ka/). 2. In voiced sounds, for /za/ and /ra/ the correct answer rate was improved by repetition or elongation of the consonant burst or transition part, and for /ga/, /ma/, and /na/ the rate was improved by amplification. 3. Semivowels (/wa, ja/) and glottal sounds (/ha/) were seldom misunderstood, and required no processing. Digital filtering processing was performed on monosyllables with the "s" sound (/sa, su, se, so/), and these filtered sounds were presented to the patients and a comparison was made with the original sounds. As a result, it was revealed that the correct answer rate could be improved by filtering, although the pass band was changed slightly by the succeeding vowel. This improvement was more apparent in the presence of environmental noise than under quiet conditions.