文章摘要
郭颖,彭任华,郑成诗,李晓东.偏度最大化多通道逆滤波语音去混响研究*[J].,2019,38(1):58-67
偏度最大化多通道逆滤波语音去混响研究*
Maximum Skewness-based Multichannel Inverse Filtering for Speech Dereverberation
投稿时间:2018-04-03  修订日期:2018-12-29
中文摘要:
      房间混响会降低语音质量和语音可懂度。高阶统计量是衡量非高斯性的重要参量,基于语音非高斯特性可实现语音去混响。本文提出一种基于高阶统计量的多通道语音去混响方法,该方法首次用多通道语音信号线性预测残差的三阶统计量偏度(Skewness)构造代价函数,以去混响重建信号线性预测残差的偏度最大化为目标自适应地更新逆滤波器;同时结合语音信号的产生模型,提出基于偏度准则的线性预测与房间脉冲响应逆滤波联合估计方法,进一步提高去混响算法性能。实验结果表明,该方法相较于已有的基于线性预测残差四阶统计量峰度(Kurtosis)的方法具有更好的去混响效果,且对噪声具有更强的鲁棒性。
英文摘要:
      Room reverberation often leads to reduce speech quality and speech intelligibility. Speech dereverberation can be achieved by using non-Gaussian property of speech, where higher order statistics (HOS) are typical measurements. This paper presents a method based on HOS for multichannel speech dereverberation. We construct the cost function using the third-order statistics, namely skewness, of multichannel speech signal linear prediction residuals, and then update the inverse filter adaptively by maximizing the skewness of the linear prediction residuals of the reconstructed speech signal; Meanwhile, by considering a speech generation model, this paper further develops a method, jointly estimating the channel’s inverse filter and the prediction error filter, which can improve the dereverberation performance. Experimental results show that the proposed method is superior to the method based on forth-order statistics, i.e. kurtosis, in terms of dereverberation and robustness to the noise.
DOI:10.11684/j.issn.1000-310X.2019.01.009
中文关键词: 高阶统计量,偏度,线性预测,房间脉冲响应,逆滤波
英文关键词: Higher order statistics (HOS), Skewness, Linear prediction (LP), Room impulse response (RIR), Inverse filtering
基金项目:国家自然科学基金项目 (61571435)
作者单位E-mail
郭颖 中国科学院声学研究所噪声与振动重点实验室 guoying@mail.ioa.ac.cn 
彭任华 中国科学院声学研究所噪声与振动重点实验室 pengrenhua@mail.ioa.ac.cn 
郑成诗 中国科学院声学研究所噪声与振动重点实验室 cszheng@mail.ioa.ac.cn 
李晓东 中国科学院声学研究所噪声与振动重点实验室 lxd@mail.ioa.ac.cn 
摘要点击次数: 23
全文下载次数: 26
查看全文   查看/发表评论  下载PDF阅读器
关闭