Simulation of Speech Enhancement Algorithms and its Performance Comparison

The main aim of the speech enhancement is to refine the human perception of system decoding by processing noisy voice signals. In the speech enhancement process, suppression of noise and enhancing the speech signals has been a prime research problem for several decades. The problem-focused in this study is enhancing the speech signal degraded by background noise. To enhance voice quality and understandability in a noisy setting, a speech enhancement algorithm must be used. This study proposes that a noisy speech signal can be enhanced by various existing algorithms such as Basic Spectral Subtraction (BSS), Over Spectral Subtraction (OSS), Maximum a posterior (MAP) estimator of magnitude and Discrete Wavelet Transform (DWT). In this paper, a simulation of the above-mentioned speech enhancement algorithms and their performance comparison analysis has been performed in terms of Signal to Noise Ratio (SNR) and Perceptual Evaluation Speech Quality (PESQ) using MATLAB.