一种自适应回声抵消和噪声消除算法

为解决回声和噪声严重影响通信系统的通信质量问题,采用带有语音活动性检测VAD(voiceactivitydetection)的谱减法消除噪声,并用基于两矢量夹角法双向通话检测器DTD(doubletalkdetector)的仿射投影算法APA(affineprojectionalgorithm)对回声自适应进行估计,提出了一种回声抵消和背景噪声消除的方法。仿真实验表明:该方法可有效地消除背景噪声,对回声路径参数的估计精确性较高,可以准确地检测到双向通话情况,同时也能够区分双向通话和回声路径突然改变这两种不同的情况,收敛速度快,且具有一定的鲁棒性。

Adaptive echo cancellation and noise reduction algorithm

Echo and noise severely impacts communication quality in communications systems. By using spectral subtraction method with voice activity detection (VAD) technique and affine projection algorithm (APA) with double talk detector (DTD), an acoustic echo and noise canceller (AENC) was proposed. Simulation results show that the proposed AENC method can efficiently reduce noise, possesses high echo parameter estimation accuracy in fast convergent speed and accurate double talk detection, and distinguishes double-talk and echo path change with robustness.