一种采用旁瓣增强的麦克风阵列抗混响算法

发布时间：2018-10-26 17:59

【摘要】：在音频/视频会议、人机交互、语音识别等领域,严重的混响干扰导致麦克风阵列语音处理性能急剧下降.针对现有逆滤波等抗混响方法需要获得准确的房间传输响应,而波束形成方法抗混响性能有限的问题,基于广义旁瓣抵消器(generalized sidelobe canceller,GSC)结构提出一种采用旁瓣增强的麦克风阵列抗混响算法(Sidelobeenhanceing reverberation mitigation algorithm for microphone array,SERM).该算法首先进行波束形成,初步获得增强的直达语音信号,并对旁瓣获取的混响分量进行自适应迭代增强,再将旁瓣迭代增强的误差信号作为参考噪声进行自适应噪声抵消,最终输出抗混响语音.实验结果表明,在混响环境下该方法能有效改善麦克风阵列的语音信号质量.
[Abstract]:In audio / video conferencing, human-computer interaction, speech recognition and other fields, severe reverberation interference leads to a sharp decline in speech processing performance of microphone arrays. In order to solve the problem that existing anti-reverberation methods such as inverse filtering need to obtain accurate room transmission response, but beamforming methods have limited anti-reverberation performance, based on generalized sidelobe canceller (generalized sidelobe canceller, the anti-reverberation performance of beamforming method is limited. GSC) A microphone array anti-reverberation algorithm (Sidelobeenhanceing reverberation mitigation algorithm for microphone array,SERM) with sidelobe enhancement is proposed. In this algorithm, the enhanced direct speech signal is initially obtained by beamforming, and the reverberation component obtained by the sidelobe is enhanced by adaptive iterative method, then the error signal enhanced by the sidelobe iteration is used as the reference noise for adaptive noise cancellation. The final output anti-reverberation voice. Experimental results show that the method can effectively improve the speech signal quality of microphone array in reverberation environment.
【作者单位】：厦门大学海洋与地球学院水声通信与海洋信息技术教育部重点实验室;厦门大学信息科学与技术学院;
【基金】：福建省高校产学合作项目(2015H6019) 福建省中青年教师教育科研项目(JA14008) 厦门市科技计划项目(3502Z20153002)
【分类号】：TN912.3

【相似文献】