基于多种子自适应分组成对比较的音质评价  

Evaluation of sound quality based on multi⁃seed adaptive grouped paired comparison

作　　者：王泽俊 林志斌 陶建成 Wang Zejun;Lin Zhibin;Tao Jiancheng(Key Laboratory of Modern Acoustics,Institute of Acoustics of Nanjing University,Nanjing,210008,China)

机构地区：[1]近代声学教育部重点实验室,南京大学声学研究所,南京210008

出　　处：《南京大学学报（自然科学版）》2021年第2期327-333,共7页Journal of Nanjing University（Natural Science）

基　　金：国家自然科学基金(11874218)。

摘　　要：声品质主观评价实验中,如果样本数量较大或已完成现有样本评价后又有新样本加入评价系统,传统的成对比较法所需的实验时间较长,难以实现,且自适应分组成对比较法缺乏对多个关联样本的研究.基于自适应分组成对比较法提出多种子自适应分组成对比较法,通过多个种子样本建立任意两组样本之间的映射关系,从而获得全体样本的评价结果.阐述了多种子自适应分组成对比较法的基本原理及具体实现方法,给出多种子自适应选取种子的经验公式.并通过设计两种主观评价实验,与传统成对比较法和自适应分组成对比较法对比,验证了所提方法的有效性和可靠性.In the subjective evaluation experiment of sound quality,when the number of samples is large or new samples need to be added to an existing completed evaluation system,the traditional paired comparison method for experiments is time consuming and unrealizable.In addition,Adaptive Grouped Paired Comparison(AGPC)method is somewhat limited by lack of relevant research on the relatedness among multiple samples.In this paper,Multi⁃seed Adaptive Grouped Paired Comparison(MAGPC)is proposed based on AGPC method.Multiple seed samples are used to establish the mapping relationship between any two sets of samples,so as to obtain the evaluation results of all samples.Firstly,the concrete implementation method of MAGPC is described.Secondly,the empirical formulas of MAGPC are given.Finally,by designing two subjective evaluation experiments,and comparing with the traditional paired comparison method and the AGPC method,the effectiveness and reliability of the proposed method are verified.

关 键 词：声品质 多种子样本 成对比较法 主观评价实验 

分 类 号：O429[理学—声学]