摘要
本文阐述了一种基于听觉模型的客观音质评价方法研究．我们在使用谱失真和谱包络失真进行客观音质评价时采用了对听觉系统模拟的三个听觉感知模型．实验结果表明：基于听觉模型的谱失真方法测量结果与主观评价更具一致性：在与主观评价具有同样相关系数时，采用听觉模型谱包络失真方法的预测系数为４阶，而ＬＰＣ倒谱距离方法的预测系数为８阶．
This paper studies briefly measures for objective quality evaluation using perceptuallybased features.We used 3 sorts of perceptuallybased features for objective quality measures of spectral distortion and spectral envelop distortion.The experimental results show that the correspondence between spectral distortion measure of perceptuallybased features and the subjective quality evaluation (MOS) is better compared with general spectral distortion measure.The predeictive order of spectral envelop distortion used perceptuallybased features is reduced from 8 to 4,and the correlation coefficients between 2 objective measures and the subjective quality are the same.
出处
《电子学报》
EI
CAS
CSCD
北大核心
1999年第7期92-94,共3页
Acta Electronica Sinica
关键词
客观音质评价
听觉模型
语音信号处理
Objective speech quality evaluation,Perceptuallybased feature,Spectral distortion measure,Spectral envelop distortion measure
