Improved speech enhancement by removal of impulsive noise

Abstract

This paper presents enhancement of speech by removal of the impulsive noise in the speech signals. The impulsive noise results in surface degradation of audio recordings and the degradation of intelligibility in speech signals. To enhance the speech component from impulsive disturbance we go for emphasis, signal segmentation and log MMSE filtering. Pre-emphasis is done to increase the magnitude of high frequencies in the speech signal. Then the signal is segmented and the segmented frames are filtered in log MMSE filter. This filter does not require an inventory of pre-recorded clean signals. This method proposes a speaker independent speech enhancement.