一种广义旁瓣抵消器结构的语音增强改进算法

发布时间：2018-05-09 12:02

本文选题：广义旁瓣抵消器 + 自适应算法　；参考：《数据采集与处理》2017年02期

【摘要】：当广义旁瓣抵消器(Generalized sidelobe canceller,GSC)结构的语音增强算法对语音信号的入射方向角估计不准确时,阻塞矩阵(Blocking matrix,BM)不能完全阻塞目标语音,使得部分语音通过阻塞矩阵,在后期多输入抵消器(Multiple-input canceller,MC)模块中和参考信号相抵消,造成目标语音的损失。针对广义旁瓣抵消器因信号到达方向(Direction of arrival,DOA)估计误差而导致语音泄漏的问题,本文提出了一种麦克风阵列语音增强的优化算法,先对经过时延补偿的信号进行频谱调整,再利用MC模块输出与BM模块输出存在相关性的特点,对阻塞矩阵进行自适应调整,使方向估计参数更趋近于真实目标语音方向,以减少阻塞矩阵中目标语音的泄漏。仿真结果表明,该算法可以有效减少阻塞矩阵中目标语音的泄漏、增强系统的鲁棒性以及提高语音增强效果。
[Abstract]:When the speech enhancement algorithm of Generalized sidelobe canceller (GSC) structure is not accurate, the blocking matrix cannot completely block the target speech, which causes some speech to pass through the blocking matrix. In the later multi-input canceller (Multiple-input canceller MCS) module, the reference signal cancels the target speech, resulting in the loss of the target speech. Aiming at the problem of speech leakage caused by the direction of arrival (DOA) estimation error of the generalized sidelobe canceller, an optimized speech enhancement algorithm for microphone array is proposed in this paper. Firstly, the frequency spectrum of the signal compensated by time delay is adjusted. Then, by taking advantage of the correlation between MC module output and BM module output, the blocking matrix is adaptively adjusted to make the direction estimation parameters closer to the real target speech direction, so as to reduce the leakage of the target speech in the blocking matrix. Simulation results show that the proposed algorithm can effectively reduce the leakage of the target speech in the blocking matrix, enhance the robustness of the system and improve the speech enhancement effect.
【作者单位】：南京邮电大学通信与信息工程学院;江苏省通信与网络技术工程研究中心;
【基金】：江苏省高校自然科学研究重大项目(13KJA510003)资助项目江苏省高校优势学科建设工程(PAPD)资助项目江苏省普通高校研究生科研创新计划(KYLX15_0851)资助项目
【分类号】：TN912.3

【相似文献】