摘要
在麦克风阵列语音增强方法中,传统的广义旁瓣抵消器在处理存在显著脉冲噪声的语音信号时效果较差。为提高在脉冲噪声干扰下的语音信号增强效果,提出一种麦克风阵列的协同自适应滤波语音增强方法。该方法采用协同自适应滤波取代线性自适应滤波,基于NLMS算法导出了滤波器权值和协同因子的自适应更新算法。仿真实验结果表明,所提方法能有效地消除掉语音信号的脉冲噪声和高斯噪声,克服线性自适应滤波对非线性脉冲噪声的不敏感性,比广义旁瓣抵消器效果优越很多。
Among the microphone array-based speech enhancement methods,the conventional generalized sidelobe canceller(GSC)has a poor effect in processing speech signals with obvious impulse noises. Therefore,a speech enhancement method with collaborative adaptive filtering and microphone array is proposed to improve the enhancement effect of speech signals under the impulse noise interference. In the method,the collaborative adaptive filtering is used to replace the linear adaptive filtering.The filter weight and co-factor adaptive updating algorithm is derived on the basis of the NLMS algorithm. The proposed method can effectively eliminate the impulse noise and Gaussian noise of speech signals,overcome the insensitivity of linear adaptive filtering to nonlinear impulse noises,and has a more superior effect than the GSC.
引文
[1]尹波.基于麦克风小阵列的语音增强算法的研究[D].武汉:湖北工业大学,2016.YIN Bo. Research on speech enhancement algorithm based on small array of microphones[D]. Wuhan:Hubei University of Technology,2016.
[2]陆振宇,何珏杉,赵为汉.关于多通道语音去噪的识别优化研究[J].计算机仿真,2016,33(6):315-320.LU Zhenyu,HE Jueshan,ZHAO Weihan. Recognition and op-timization of denoising method for multi-channel speech[J].Computer simulation,2016,33(6):315-320.
[3]陈磊,江伟华,童峰,等.一种可跟踪移动声源方向的麦克风阵列语音增强算法[J].厦门大学学报(自然科学版),2015,54(4):551-555.CHEN Lei,JIANG Weihua,TONG Feng,et al. A microphone array speech enhancement algorithm with tracking capability upon moving speakers[J]. Journal of Xiamen University(Natu-ral science),2015,54(4):551-555.
[4]高金.基于麦克风阵列的移动声源定位研究[D].秦皇岛:燕山大学,2015.GAO Jin. Research on mobile sound source localization based on microphone array[D]. Qinhuangdao:Yanshan University,2015.
[5]高媛.基于最小均方误差的稀疏自适应滤波算法研究[D].沈阳:沈阳工业大学,2014.GAO Yuan. Research on sparse adaptive filtering algorithm based on minimum mean square error[D]. Shenyang:Shenyang University of Technology,2014.
[6]徐进,赵益波,郭业才.一种新的麦克风阵列自适应语音增强方法[J].应用科学学报,2015,33(2):187-193.XU Jin,ZHAO Yibo,GUO Yecai. A new adaptive speech en-hancement method for microphone array[J]. Journal of applied sciences,2015,33(2):187-193.
[7]郑璟瑜.基于改进小波阈值去噪的无模型自适应控制抗噪方法[D].北京:北京交通大学,2016.ZHENG Jingyu. Model-free adaptive control anti-noise method based on improved wavelet threshold denoising[D]. Beijing:Beijing Jiaotong University,2016.
[8]赵益波,徐进,阮玮琪,等.一种改进GSC自适应语音增强方法[J].科技通报,2016,32(9):172-176.ZHAO Yibo,XU Jin,RUAN Weiqi,et al. An improved GSC adaptive speech enhancement method[J]. Bulletin of science and technology,2016,32(9):172-176.
[9] COMMINIELLO D,SCARPINITI M,AZPICUETA-RUIZ L A,et al. Combined nonlinear filtering architectures involving sparse functional link adaptive filters[J]. Signal processing,2017,135:168-178.
[10] CARINI A,CECCHI S,ROMOLI L,et al. Legendre nonlin-ear filters[J]. Signal processing,2015,109:84-94.
[11]肖累累.语音质量客观评价方法的研究[D].北京:北京邮电大学,2014.XIAO Leilei. Research on objective evaluation method of speech quality[D]. Beijing:Beijing University of Posts and Telecommunications,2014.
[12]张伟伟.通信系统中语音质量评价的研究[D].北京:北京邮电大学,2014.ZHANG Weiwei. Research on speech quality evaluation in communication system[D]. Beijing:Beijing University of Posts and Telecommunications,2014.