VAD system under uncontrolled environment: A solution for strengthening the noise robustness using MMSE-SPZC

Voice activity detection (VAD) plays a crucial role in speech processing, serving as a fundamental component for various applications such as speech recognition and communication systems. Numerous approaches have been explored to address the VAD issue, but their effectiveness diminishes significantly in the presence of noise. In response to this challenge, we present a novel technique designed to enhance the robustness of VAD under noisy conditions. Our proposed system incorporates a background noise suppression module based on the minimum mean square error spectrum power estimator using zero crossing (MMSE-SPZC). This module is integrated before the semi-supervised Gaussian mixture model-based VAD (SSGMM-VAD). The effectiveness of the proposed VAD system is evaluated across four distinct types of noise and low signal-to-noise ratio (SNR) levels. Our experimental results reveal significant improvements in VAD accuracy, demonstrating the system’s ability to maintain robust performance even in the presence of challenging background noise conditions.