语音质量客观评价的一步策略  

One-Step Strategy of Speech Quality Objective Assessment

作　　者：付强[1] 易克初[1] 田斌[1] 张知易[2] 

机构地区：[1]西安电子科技大学ISN国家重点实验室,陕西西安710071 [2]西南通信研究所,四川成都610041

出　　处：《电子学报》2001年第7期885-887,共3页Acta Electronica Sinica

基　　金：国家自然科学基金 (No .69872 0 2 7)

摘　　要：本文提出了一种基于一步策略的语音质量客观评价新方法 .它利用前向神经网络的多维非线性映射原理 ,将传统的包含平均失真测度计算和平均失真测度到MOS的映射这两个步骤合为一步来实现 .其特点是可以充分地反映人耳听觉系统的感知特性 ,且计算简便 .在统计学意义上它还是MOS的一致估计 .主客观评价结果的相关性实验表明 ,一步法的相关系数可达到 0 95左右 。This paper proposes a novel method for objective assessment of speech quality based on one-step strategy. Using the theory of multi-dimension non-linear mapping of the feedforward neutral network, the method combines the traditional two steps of average distortion computing and mapping from the average distortion value to the Mean Opinion Score (MOS) into one step. It can embody adequately the perception properties of the human auditory system with simple computing and also is a unanimous estimate for MOS in statistical sense. Experimental results show that the correlation coefficient between the subjective test score and objective MOS estimate of one-step method can reach up to about 0.95, which is obviously better than that of the conventional two-step methods.

关 键 词：语音质量 客观评价 神经网络 

分 类 号：TN912.3[电子电信—通信与信息系统]