论文部分内容阅读
语义细分法是声品质以及声感知特征研究中最广泛采用的主观评价方法。但对于一些声品质指标,尤其是新提出的指标或评价者认识和观点不完全一致的指标,由于评价者尺度掌握的困难,采用传统语义细分法评价难以获得准确可靠的评价结果。为此,提出了采用参考激励信号的参考语义细分法评价程序,并研究了这种评价程序的性能表现。分析比较了两种语义细分法在评价者的个体重复性、团组一致性以及与成对比较法和计算结果的相关性。研究结果显示,对传统语义细分法难以获得稳定可靠评价结果的参量,参考语义细分法可以显著改善评价实验性能;对传统语义细分法可以获得稳定可靠评价结果的参量,参考语义细分法可以获得一致的评价结果,且改善了一些方面的性能。研究结果证明了采用参考激励信号的参考语义细分法的优越性。
Semantic subdivision is the most widely used method of subjective evaluation in the study of acoustic quality and acoustic perception. However, some of the indicators of sound quality, especially the newly proposed indicators or evaluators are not fully consistent with the understanding of indicators and indicators, due to the difficulty of evaluation criteria, the use of traditional semantic subdivision evaluation is difficult to obtain accurate and reliable evaluation results. For this reason, a reference semantic subdivision evaluation program using reference excitation signal is proposed and the performance of this evaluation program is studied. Analysis and comparison of two semantic subdivisions in the evaluators of individual repeatability, group consistency and paired comparison method and the results of the correlation. The results show that the reference semantic subdivision method can significantly improve the performance of the evaluation experiment when the traditional semantic subdivision method is difficult to obtain a stable and reliable evaluation result. For the traditional semantic subdivision method, the parameters of the stable and reliable evaluation result can be obtained, and the reference semantic subdivision Law can get consistent evaluation results, and improve some aspects of performance. The results of the study demonstrate the superiority of the reference semantic subdivision using a reference excitation signal.