An effective voice activity detection algorithm in mobile communication corrupted by impulse noise

Existing voice activity detection (VAD) algorithms degrade drastically in mobile communication environments corrupted by impulse noise. A novel fusion algorithm is proposed for VAD adapted to such environments. Theoretical and experimental analyses were done to justify the effectiveness of the algorithm, which demonstrated that the proposed VAD algorithm operates accurately and reliably at the presence of impulse noise even with a varying signal-to-impulse-noise ratio (SINR). Moreover the proposed algorithm needs less computational complexity overhead and is suitable for real-time implementation.