A hybrid approach for speech enhancement using Bionic wavelet transform and Butterworth filter

Abstract

Digital noise reduction schemes are proposed for minimizing the output of the hearing aid in the existence of noise. That oversimplification does not observe the reality that individual perception and preference determine which sounds are required and which sounds are not required. Though numerous approaches have been endorsed to diminish the noise level, there is a necessity of development in the noise reduction systems. This paper offered a hybrid method to shrink the commotion in the signals. First, the robust signal is fetched from the database and that signal is subjected to pre-reconstruction with the assistance of empirical mode decomposition. After that the input to Bionic wavelet transform (BWT) is set as the resultant reconstructed signal to derive the de-noised signal. Butterworth filter is used for fine tuning de-noised signal. Consequently, inverse BWT is utilized to derive the final de-noised signal. The recommended methodology is implemented in the MATLAB platform. This method is assessed utilizing more number of noises; furthermore, the technique's performance is matched with the prevailing noise removal techniques.