论文部分内容阅读
空间目标语音增强技术在移动语音应用中具有重要价值。为有效抑制空间干扰语音和背景噪声,采用声学矢量传感器(acoustic vector sensor,AVS),利用其结构的特殊性和各输出分量间的三角函数关系,引入阻塞矩阵,推导目标语音信号的功率谱,从而实现对AVS全向分量的最佳Wiener滤波获得增强语音。仿真实验表明,该方法能够有效抑制空间语音干扰和背景噪声。该方法采用单AVS,计算复杂度低,具备移动设备上的应用优势。
Spatial target speech enhancement technology is of great value in mobile voice applications. In order to effectively suppress the spatial interference of speech and background noise, an acoustic vector sensor (AVS) is used to introduce the blocking matrix to derive the power spectrum of the target speech signal by using the particularity of its structure and the trigonometric functions among the output components. In this way, the optimal Wiener filtering for the AVS omnidirectional components achieves enhanced speech. Simulation results show that this method can effectively suppress spatial speech interference and background noise. The method uses a single AVS, low computational complexity, with the advantages of mobile devices.