On some speech enhancement strategies for improved perception by the hearing impaired

A promising scheme for enhancing the speech signal is based on studies of speaking clearly for the hearing Impaired. Two characteristics of clear (intelligible) speech, namely consonant-to-vowel Intensity ratio (CVR) and consonant duration (CD), have been chosen for evaluation In the present Investigation. The aim of the experiments reported here was to study the extent of CVR modification and CD modification that would be helpful in speech perception by the hearing Impaired. A computerised test administration was developed In order to automate the experiments. It was decided to use synthetic stimuli since the segmentation of individual phonemes would be accurate, and it would be possible to manipulate the various acoustic segments Independently. The stimuli Involved the English stop consonants In the /CV/ and /VC/ contexts of various vowels. The subject were normal-hearing Impairment was simulated by mixing each stimulus with masking broadband noise. The results of each experimental run were obtained In the form of a stimulus-response confusion matrix. The confusion matrices were evaluated on the basis of recognition scores and Information transmission anlaysis. The result show that Increasing CVR does Improve recognition scores, The Information transmitted on the basis of feature-classification, as well as the average response times were also found to Improve with increasing CVR. The results for the OD modification experiments suggest that at higher noise levels, a formant transition duration modification of up to about 50% may be combined with burst duration modification and expected to yield better performance. The above results suggest that it would be fruitful to develop a scheme that can identify the boundaries of the various phonemic segments and subsegments and then perform appropriate modifications In C/V ratio, formant transition duration, and burst duration. This scheme could then be tested on the hearing impaired.