声品质主观评价的分组成对比较法研究

声品质主观评价实验中,大样本量的成对比较法评价实验由于评价时间随样本量的平方增长而变得难以实现,为此提出了可有效减少评价工作量并使得实验易于实现的分组成对比较法。将总体评价样本分成具有相互联系的n个样本组分别进行主观评价,通过各组间的关联样本将各组的评价结果进行反演构建,得到总体样本的评价结果,进而分析其声品质特征。以车内噪声为例,介绍了分组成对比较法的实现方法,并与传统成对比较法以及语义细分法实验结果进行了对比分析,两者结果获得较好的一致性,证明了分组成对比较主观评价实验方法的有效性。

Grouped pair-wise comparison for subjective sound quality evaluation

In subjective sound quality pair-wise comparison evaluation, time consumption grows with square of number of sound stimuli, which made the evaluation of large quantity of stimuli difficult to carry out. A grouped pair-wise comparison design (GPC) is proposed to greatly decrease time requirement and difficulties of subjective comparison test, where the whole evaluation stimuli are divided into n test groups, with reference link stimuli among each group. From subjective results of each group, final results of all stimuli are reconstructed, and their sound quality can be analyzed. With car interior sound stimuli as example, realization of the proposed GPC method is introduced. The results of GPC method are in good agreement with those obtained from pair-wise comparison and semantic differential method.