Leeuw A R, Dreschler W A
Department of Clinical Audiology, University of Amsterdam, The Netherlands.
Ear Hear. 1994 Jun;15(3):240-55. doi: 10.1097/00003446-199406000-00005.
Knowledge of the frequency resolving power of the pathological ear can be helpful for good hearing-aid prescription. In particular, the slope of the low-frequency edge of the auditory filter will determine the extent of upward spread of masking. The most precise method for measuring this filter is the "notched-noise" procedure developed in the early 1980s. However, at this moment, the notched-noise protocol is too time consuming for clinical use. In this paper we developed a protocol that is applicable in the clinic. In the first part of the study we investigated the minimum number of threshold measurements necessary for a reliable estimation of the filter parameters. For this purpose we analyzed 99 filters originally obtained with 13 threshold measurements. The influence of reducing the number of notch widths used to obtain filter shapes on the variability of filter parameters was investigated. Subsets of seven and five notch widths gave about the same standard deviations of the differences between the parameters obtained with the subset and the parameters obtained with the full set of 13 thresholds, while subsets of four and three notch widths gave significantly higher variability. However, because small deviations of one threshold determination can give rise to a large change in filter parameters, especially for filters with flatter skirts, it is strongly recommended to reduce the variability by measuring thresholds twice (test and retest). In the second part of the study we investigated a faster method for measuring thresholds. The forced-choice paradigm normally used in notched-noise procedures was replaced by a Békésy paradigm. The Békésy paradigm did not significantly increase intra-individual standard deviations, but did reduce the measuring time by more than 50%. In conclusion, the new procedures reduce the measurement time needed to obtain reliable estimates of auditory-filter parameters by a factor of about 5. The new protocol lasts about 15 min for each filter measurement, which appears to be acceptable for clinical use, at least for difficult-to-fit hearing aid users.
了解患耳的频率分辨能力有助于进行良好的助听器验配。特别是，听觉滤波器低频边缘的斜率将决定掩蔽向上扩展的程度。测量该滤波器最精确的方法是20世纪80年代初开发的“带缺口噪声”程序。然而，目前，带缺口噪声方案在临床使用中耗时过长。在本文中，我们开发了一种适用于临床的方案。在研究的第一部分，我们调查了可靠估计滤波器参数所需的最小阈值测量次数。为此，我们分析了最初通过13次阈值测量获得的99个滤波器。研究了减少用于获得滤波器形状的缺口宽度数量对滤波器参数变异性的影响。七个和五个缺口宽度的子集给出的子集参数与完整的13个阈值集获得的参数之间差异的标准差大致相同，而四个和三个缺口宽度的子集给出的变异性明显更高。然而，由于一次阈值测定的小偏差可能会导致滤波器参数的大幅变化，特别是对于裙边较平坦的滤波器，强烈建议通过测量两次阈值（测试和重新测试）来降低变异性。在研究的第二部分，我们研究了一种更快的阈值测量方法。通常用于带缺口噪声程序的强制选择范式被贝凯西范式所取代。贝凯西范式没有显著增加个体内标准差，但确实将测量时间减少了50%以上。总之，新程序将获得听觉滤波器参数可靠估计所需的测量时间减少了约5倍。每个滤波器测量的新方案持续约15分钟，这对于临床使用似乎是可以接受的，至少对于难以适配助听器的用户来说是这样。
