A VAD-based switch fast LMS/Newton algorithm for acoustic echo cancellation

This paper studies a new switch fast LMS/Newton algorithm for acoustic echo cancellation. By employing the envelop tracking voice activity detection technique, an efficient switch scheme is developed, which can help the proposed algorithm distinguish the significant and insignificant input speech periods and thus work between the fast LMS/Newton and NLMS modes, respectively. Hence, convergence can be improved when the input excitation is sufficient and the computational complexity can be reduced when the input is relatively weak. Finally, simulations are conducted to prove the efficiency of the proposed algorithm.