基于GSC结构的多麦克风数字助听器的自适应波束形成算法

针对利用麦克风阵列进行数字助听器的语音增强这一应用,本文对传统的GSC结构的自适应波束形成算法加以改进,以对固定波束形成器(fixed beamfomer,FBF)的输出进行语音活动检测的结果,来取代信干比作为处理过程中的自适应模式控制,并将波束形成过程对目标语音信号的畸变引入到滤波器代价函数的设计,重新定义多路输入抵消器(multiple-input canceller,MC)中的归一化最小均方算法的迭代公式,以减少自适应过程中目标语音的损失.仿真结果显示使用新的代价函数后,背景噪声和强干扰得到了明显的抑制,目标语音信号的畸变显著减小,而算法因此增加的计算量也在一个可接受的程度.

An adaptive Beamforming Algorithm in GSC Structure for Multi-Microphone Digital Hearing Aids

Some modifications were made to traditional adaptive GSC beamformer for application to speech enhancement in digital hearing aids.The voice activity detecting(VAD) result of fixed beamformer(FBF) output instead of signal to intecference ratio(SIR) served as the adaptation mode control during the process,and speech distortion caused by the beamforming process was taken into the consideration of cost function designed for the adaptive filter.As a result,the iterative formula was redefined in Multiple-input Canceller so as to decrease the speech cancellation.The simulation experiments exhibited drastic reduction of background noise and strong interferences,along with obviously less speech distortion after the adoption of the new cost function.However the amount of the calculation was kept at an acceptable level.