SPEECH ENHANCEMENT USING CASCADED SCHEMES OF PHASE SPECTRUM COMPENSATION (PSC) AND MODIFIED SPECTRAL SUBTRACTION

Abstract

The purpose of this exploratory article is to offer a novel method for improving speech signals in noisy environments by utilising protruded systems of phase spectral compensation (PSC) and modified spectral deduction. In order to improve the quality of speech, the method that has been presented would handle both the spectral and phase deformations that are caused by external noise. Initially, phase spectral compensation is utilised in order to rectify phase deformations that are brought about by noise. This is then followed by modified spectral deduction in order to successfully suppress noise factors. By virtue of their protruding nature, these schemes guarantee a comprehensive decrease in noise while still preserving the intelligibility and lightheartedness of speech. The results of the experiments show that there have been considerable advancements in speech quality criteria, such as the signal-to-noise ratio (SNR) and the perceptual evaluation of speech quality (PESQ), across a variety of different types and conditions including colourful noise. The findings of this investigation provide a potential framework for real-world operations that are utilising robust speech improvement methods in noisy environments.