Moore响度计算模型的改进

A revision of Moore's loudness model

与早期响度的图表计算方法相比,Moore计算模型是基于解析式的并且实现了响度值随频率、强度改变的连续计算,因此使用起来非常方便;然而,Moore模型计算公式中基本参数值的确定所依据的闻阈标准数据已被修订,并且最新研究表明,Moore模型在确定α值时所做的假设是不合理的;因此本文中(1)提出了利用耳蜗输入输出函数以确定响度函数的方法;(2)根据多个频率处的响度函数重新估计了Moore模型的参数α;(3)根据最新的闻阈标准数据和新的机理重新修订了其它基本参数;(4)利用新的模型重新计算了等响曲线并与原有模型的结果以及最新国际标准进行了对比。结论表明对1～8kHz之间的计算结果有了较大的修订,特别是能很好地预测1～2kHz之间的凸起。

Comparing with earlier loudness＇ models, Moore＇ s model has the advantage that loudness＇ computation is based on analytical formulae rather than by reference to charts or tables. However, parameters in these analytical formulae must be revised because the absolute thresholds have been revised, and according to the latest research, several suppositions in Moore＇s model in determining α are questionable. In this paper, the following steps are taken ： （ 1 ） put forth a way to acquire loudness functions from cochlear input-output functions ; （2） estimate parameter α from the loudness function at 0.25, 0.5, l, 2, 3, 4, 6 and 8 kHz; （3） revise other parameters so as to be based on new data and modified approaches ; （4） compare equal-loudness-level contours of revised model with the old model and ISO-226 （2003）. The result shows the loudness in the frequency range between 1 to 8 kHz is greatly improved, especially, the peak in the frequency range between 1 to 2 kHz can be predicted.