Abstract
Human speech is the principal medium of communication in modern telecommunication systems, enabling efficient human-to-human and human-to-machine interactions. However, speech signals transmitted through communication channels are often degraded by additive white Gaussian noise (AWGN), leading to reduced speech quality and intelligibility. This study presents a novel Blackman Window-Based Shrinkage Soft Thresholding (Bman-WSST) technique for enhancing speech signals contaminated by AWGN. The proposed method integrates a newly developed Blackman window-based shrinkage function with the Discrete Wavelet Transform (DWT) to improve noise suppression while preserving essential speech characteristics. A speech signal comprising 73,113 samples, sampled at 44.1 kHz, was corrupted with 10.5 dB AWGN to evaluate the performance of the proposed algorithm. Speech denoising was performed using the Symlet 4 (Sym4) mother wavelet at decomposition levels 1–4 and assessed using four threshold selection rules: Sqtwolog, Rigrsure, Heursure, and Minimax. The results identified the Sqtwolog threshold selection rule and Level 2 decomposition as the optimum configuration. Performance evaluation based on the power spectral density (PSD), signal-to-noise ratio (SNR), mean square error (MSE), and maximum absolute error (MAE) yielded values of −68.23 dB, 25.553 dB, 0.0108, and 0.07870, respectively. These findings demonstrate that the proposed Bman-WSST technique effectively suppresses AWGN while preserving speech fidelity, providing a robust and efficient alternative to conventional wavelet coefficient shrinkage methods for speech enhancement in telecommunication systems.

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Copyright (c) 2026 Scholarly Journal of Science and Technology Research and Development