Spectral Splitting of Speech by Wavelet Packets to Shrink Simultaneous Masking

Simultaneous masking is when a sound is made inaudible by a masker, a noise or unwanted sound of the same duration as the original sound. An innovative approach is investigated for speech processing in cochlea. Differently from the traditional filter-bank spectral analysis strategies, the proposed method analyses the speech signal by means of the wavelet packets. Splitting the speech signal by filtering and down sampling at each decomposition level, using wavelet packets with different wavelet functions, helps to shrink the effect of simultaneous masking. The performance of the proposed method is experimentally evaluated with vowel-consonant-vowel -syllables for fifteen English consonants. The dichotic presentation of processed speech signals effectively reduces the simultaneous masking through which it improves auditory perception.