Combined Effect of Adaptive Wiener Filter and Spectral Splitting of Speech Signal in Improving Speech Intelligibility for Hearing Impaired People
TL;DR
Results suggest that processing with a composite scheme improved speech intelligibility for hearing-impaired individuals in noisy environments.
Abstract
—The goal of this research is to look into the combined effect of an adaptive Wiener filter, frequency splitting, and amplitude compression on speech intelligibility for hearing impaired people in adverse listening environments. The adaptive Wiener filter is based on the adaptation of the filter frequency response from sample to sample depending on the voice signal statistics (mean and variance).The filter bank approach was modified to perform spectral splitting of the voice signal in order to lessen the influence of spectral masking. To address the issue of hearing loss, amplitude compression with a compression factor of 0.6 was used. As a result, the combined strategy aids in enhancing speech intelligibility for hearing-impaired individuals in noisy environments. To measure the intelligibility of processed speech, hearing-impaired subjects (moderate SNHL) were given the modified rhyming Test (MRT). The test procedure consists of 300 phrases. Each phrase is composed of consonant-vowel-consonant words (CVC). When compared to an adaptive Wiener filter, the results suggest that processing with a composite scheme improved speech intelligibility. With a lower SNR value of -6 dB, a maximum improvement in speech recognition score of 32.935 % was recorded.