Mel-Spectral Distortion Measure Based on Perception Model for Objective Speech Quality Assessment
-
摘要: 为了有效评价通信系统中的语音质量,基于语音感知分析,提出了M el域上一种新的语音信号特征表示方法———MFSC(美尔谱系数).MFSC既考虑人耳对频率的非线性感知特性,又结合了声音强度-响度非线性变换特性,符合语音感知分析.基于MFSC特征参数的提取,提出了用于语音质量客观评价的美尔谱失真测度(M el-SD),并将其应用于干扰条件下的无线通信系统语音质量评价.实验结果表明,M el-SD的平均相关值为0.942,分别比M el-CD和PESQ(语音质量感知评价)提高了0.089和0.031.Abstract: In order to assess speech quality effectively,a new approach of feature extraction of speech signals,MFSC(Mel-frequency spectral coefficient),was proposed on the basis of the speech perception model.This approach considers both the perceptual nonlinear relation between sound intensity and loudness and the nonlinear perception for frequency in Mel domain.Furthermore,an objective speech quality measure with MFSC feature parameters,referred to as Mel-SD(Mel-spectral distortion measure),was put forward,and speech quality assessment based on Mel-SD for jammed wireless communication systems was given.The experimental results show that the average correlation using Mel-SD is 0.942,increasing by 0.089 and 0.031 respectively compared with Mel-CD(Mel-cepstral distance measure) and PESQ(perceptual evaluation of speech quality).
-
陈国,胡修林,张蕴玉,等.语音质量客观评价方法研究进展[J]. 电子学报,2001,29(4):548-552.CHEN Guo,HU Xiulin,ZHANG Yunyu,et al.Research advance on objective measures of speech quality[J]. Acta Electronica Sinica,2001,29(4):548-552.[2] RIX A W.Perceptual speech quality assessment-a review[C] // Proc.IEEE International Conference on Acoustics,Speech,and Signal Processing.Piscataway:IEEE Press,2004,3:1 056-1 059.[3] BARRETT P A,RIX A W.Applications of speech quality measurement for 3G[C] // Third International Conference on Mobile Communication Technologies.London:IEE Press,2002:250-255.[4] KIM D S.ANIQUE:an auditory model for single-ended speech quality estimation[J]. IEEE Trans.on Speech and Audio Processing,2005,13(5):821-831.[5] 杨震,毕厚杰.一种新的用于语音主观质量评价的谱失真参数[J]. 电子与信息学报,2001,23(7):669-676.YANG Zhen,BI Houjie.A new parameter of spectral distortion for predicting subjective quality of speech[J]. Journal of Electronics and Information Technology,2001,23(7):669-676.[6] ITU-T Rec.P.861.Objective quality measurement of telephone-band speech codecs[S]. 1996.[7] ITU-T Rec.P.862.Perceptual Evaluation of Speech Quality (PESQ) an objective method for end-to-end speech quality assessment of narrowband telephone networks and speech codecs[S]. 2001.[8] KUBICHEK R.Mel-cepstral distance measure for objective speech quality assessment[C] //Proc.IEEE Pacific Rim Conference on Communications,Computers,and Signal Processing.Piscataway:IEEE Press,1993:125-128.[9] 马大猷,沈壕.声学手册[M]. 北京:科学出版社,2004:570-571.MA Dayou,SHEN Hao.Handbook of acoustics[M]. Beijing:Science Press,2004:570-571.[10] DAVIS S B,MERMELSTEIN P.Comparison of parametric representations for monosyllabic word recognition in continuously spoken sentences[J]. IEEE Trans.on Acoust,Speech,Signal Processing,1980,28 (4):357-336.[11] JANKOWSKI C R J,VO H D H,LIPPMANN R P.A comparison of signal processing front ends for automatic word recognition[J]. IEEE Trans.on Speech Audio Processing,1995,3 (4):286-293.[12] 黄惠明,王瑛,赵思伟,等.语音系统客观音质评价研究[J]. 电子学报,2000,28(4):112-114.HUANG Huiming,WANG Yin,ZHAO Siwei,et al.Study of objective quality evaluation for the speech systems[J]. Acta Electronica Sinica,2001,28(4):112-114.[13] 付强,田斌,张知易,等.基于神经网络的语音谱失真测度研究[J]. 声学学报,2001,26(2):180-184.FU Qiang,TIAN Bin,ZHANG Zhiyi,et al.Research on speech distortion measure based on neural network[J]. Acta Acustica,2001,26(2):180-184.[14] 付强,易克初,田斌,等.语音质量客观评价的一步策略[J]. 电子学报,2001,29(7):885-887.FU Qiang,YI Kechu,TIAN Bin,et al.One-step strategy of speech quality objective assessment[J]. Acta Electronica Sinica,2001,29(7):885-887.[15] 赵力.语音信号处理[M]. 北京:机械工业出版社,2003:54.ZHAO Li.Speech signal processing[M]. Beijing:China Machine Press,2003:54.[16] 梁之安.听觉感受和辨别的神经机制[M]. 上海:上海科学教育出版社,1999:50.LIANG Zhian.Neural mechanisms of auditory perception and discrimination[M]. Shanghai:Shanghai Science Education Press,1999:50.
点击查看大图
计量
- 文章访问数: 1473
- HTML全文浏览量: 78
- PDF下载量: 378
- 被引次数: 0