An Impulse Noise Robust Voice Activity Detection Algorithm Applied for Low Signal-to-Noise Ratio Digital Communication

Traditional voice activity detection (VAD) algorithms deteriorate drastically in low signal-to-noise ratio (SNR) environments especially at the presence of impulse noise. This paper presents a novel algorithm based on the fusion of SNR features of selected sub-bands. The proposed algorithm was evaluated in different environments, yielding results which demonstrate that the new method can significantly improve the accuracy with impulse noise even in a signal-to-impulse-noise ratio (SINR) varying environment while maintaining high performance with other kinds of noises. Furthermore the algorithm is of less computational complexity and suitable for real-time implementation