基于时间递归平均的语音噪声功率谱估计算法研究
|
摘要
提出一种改进的时间递归平均噪声功率谱估计算法;利用谱熵计算当前语音存在概率并获取平滑系数,采用双平滑系数估计平滑后的当前语音存在概率,最后得到噪声功率谱;该算法采用自适应跟踪可以通过参数及时跟踪噪声变化,使得估计的噪声信号与原噪声信号基本保持一致;实验仿真结果证明该算法估计的噪声明显改善了时间递归平均算法估计滞后的问题,同时该算法的归一化均方误差也低于时间递归平均算法。
This paper proposed an improved time-recursive averaging power spectral estimation algorithm for non-stationary signal tracking. Firstly,the current speech presence probability was estimated by spectral entropy,then the smoothing coefficient was determined by using the speech presence probability.Finally,the noise power spectrum was obtained by using the double smoothing coefficient to estimate the smoothed current speech presence probability. In this algorithm,adaptive tracking was used to track the change of noise in time by parameters,so that the estimated noise signal was basically consistent with the original noise signal. The simulation results show that the proposed algorithm can significantly improve the estimation of time-recursive averaging algorithm. The normalized mean square error of the proposed algorithm was also lower than that of time recursive averaging algorithm.
This paper proposed an improved time-recursive averaging power spectral estimation algorithm for non-stationary signal tracking. Firstly,the current speech presence probability was estimated by spectral entropy,then the smoothing coefficient was determined by using the speech presence probability.Finally,the noise power spectrum was obtained by using the double smoothing coefficient to estimate the smoothed current speech presence probability. In this algorithm,adaptive tracking was used to track the change of noise in time by parameters,so that the estimated noise signal was basically consistent with the original noise signal. The simulation results show that the proposed algorithm can significantly improve the estimation of time-recursive averaging algorithm. The normalized mean square error of the proposed algorithm was also lower than that of time recursive averaging algorithm.
引文
[1]徐岩,王春丽.语音信号增强技术及其应用[M].北京:科学出版社,2014.
[2] KOBATAKE H. Optimization of Voice/Unvoiced Decisions in Nonstationary Noise Environments[J]. IEEE Transactions on Acoustics,Speech and Signal Processing,1987,35(1):9-18.
[3] MARTIN R. Spectral subtraction based on minimum statistic[J].Proceedings of the Seventh Europen Signal Processing Conference,EUSIPCO-94,Edingburgh,Scotland,1994(12):1182-1185.
[4] MARTIN R. Noise Power Spectral Density Estimation Based on Optimal Smoothing and Minimum Statistics[J]. IEEE Trans. on Speech and Audio Processing,2001,9(5):504-512.
[5] COHEN I,BERDUGO B. Noise estimation by minima controlled recursive averaging for robust speech enhancement[J]. IEEE Signal Proc. Letter,2002,9(1):12-15.
[6] COHEN I. Noise Spectral Estimation in Adverse Environments:Improved Minima Controlled Recursive Averaging[J]. IEEE Transactions on Speech and Audio Processing,2003,11(5):466-475.
[7] RACHAGARI S,LOIZOU P C,A noise-estimation algorithm for highly non-stationary environment[J]. Speech Communication,2006,48(2):220-231.
[8]樊新海,孙国强,张传清.基于噪声信号EMD的装甲车辆识别研究[J].兵器装备工程学报,2017,38(7):111-115.
[9] HARALD GUSTAFSSON,SVEN ERIK,INGVAR CLAESSON. Spectral subtraction using reduced delay convolution and adaptive averaging[J]. IEEE Transactions on Speech and Audio Processing,2001,10:799-807.
[10]李光源.高效语音增强与端点检测技术研究[D].北京:清华大学,2011.
[11]赵彦平,赵晓晖,王波.基于语音存在概率的噪声功率谱估计[J].吉林大学学报(工学版),2016,46(3):917-921.
[2] KOBATAKE H. Optimization of Voice/Unvoiced Decisions in Nonstationary Noise Environments[J]. IEEE Transactions on Acoustics,Speech and Signal Processing,1987,35(1):9-18.
[3] MARTIN R. Spectral subtraction based on minimum statistic[J].Proceedings of the Seventh Europen Signal Processing Conference,EUSIPCO-94,Edingburgh,Scotland,1994(12):1182-1185.
[4] MARTIN R. Noise Power Spectral Density Estimation Based on Optimal Smoothing and Minimum Statistics[J]. IEEE Trans. on Speech and Audio Processing,2001,9(5):504-512.
[5] COHEN I,BERDUGO B. Noise estimation by minima controlled recursive averaging for robust speech enhancement[J]. IEEE Signal Proc. Letter,2002,9(1):12-15.
[6] COHEN I. Noise Spectral Estimation in Adverse Environments:Improved Minima Controlled Recursive Averaging[J]. IEEE Transactions on Speech and Audio Processing,2003,11(5):466-475.
[7] RACHAGARI S,LOIZOU P C,A noise-estimation algorithm for highly non-stationary environment[J]. Speech Communication,2006,48(2):220-231.
[8]樊新海,孙国强,张传清.基于噪声信号EMD的装甲车辆识别研究[J].兵器装备工程学报,2017,38(7):111-115.
[9] HARALD GUSTAFSSON,SVEN ERIK,INGVAR CLAESSON. Spectral subtraction using reduced delay convolution and adaptive averaging[J]. IEEE Transactions on Speech and Audio Processing,2001,10:799-807.
[10]李光源.高效语音增强与端点检测技术研究[D].北京:清华大学,2011.
[11]赵彦平,赵晓晖,王波.基于语音存在概率的噪声功率谱估计[J].吉林大学学报(工学版),2016,46(3):917-921.