论文部分内容阅读
提出了一种噪声与共振环境下的数字助听器鲁棒声源定位方法,该方法基于房间共振模型和多通道自适应特征值分解算法,利用自适应次梯度投影方法迭代地估计声源至各麦克风的脉冲响应系数,从而获得各麦克风对的时延差,并利用几何方法计算声源位置。针对眼镜数字助听器四元麦克风方阵的仿真表明,所提出的自适应次梯度投影方法在强噪声环境中的鲁棒性和定位精度优于传统NLMS算法。
A robust source localization method for digital hearing aids based on room resonance model and multi-channel adaptive eigenvalue decomposition is proposed. This method uses adaptive sub-gradient projection to iteratively estimate the sound source to each microphone Of the impulse response coefficient, so as to obtain the delay difference of each microphone pair, and use the geometric method to calculate the sound source position. Simulation results show that the proposed adaptive sub-gradient projection method is superior to the traditional NLMS algorithm in robustness and positioning accuracy in strong noisy environments.