A novel double-talk detector based on adaptive estimation of near-end power

Double talk detector is the essential part of an echo canceller to keep the adaptive filter from divergence in presence of near-end speech. This work is a contribution to propose a new double talk algorithm based on power, this method detects double-talk quickly; also it can distinguish the double-talking when the channel has the varying gain. In this paper, we first present an analytical discussion describing this new method and then we show the simulation result to compare our proposed algorithm in both performance and complexity with Giegel algorithm-the most successful one in line echo cancellers.