基于熵的语音音质评价方法
Speech quality evaluation based on information entropy
-
摘要: 研究一种基于熵的语音音质客观评价方法,解决语音编、解码,通信设备研制以及通信干扰性能判定过程中长期使用主观评价的不便性。首先详细分析了信息的统计特性,在此基础上引入了马尔可夫信源并论证了语音信号的马尔可夫特性,介绍了一种用马尔可夫熵来表征语音音质变化的方法,试验数据验证了该方法具有一定的可靠性、可行性和应用前景。Abstract: A new scheme of speech quality evaluation based on information entropy is presented for design of coding-decode and communication equipment, and characterizing anti-jamming capability. This paper introduces Markov information and characteristics of speech, and measures Markov information entropy of dissimilarity speech quality. This technique was verified to be reliable, feasible and valuable.
下载: