Widex Office of Research in Clinical Amplification (ORCA), Widex Hearing Aid Company, Lisle, Illinois 60532, USA.
Ear Hear. 2010 Dec;31(6):779-95. doi: 10.1097/AUD.0b013e3181e97bfb.
Many new processing features in hearing aids have their primary effects on information located in the high frequencies. Speech perception tests that are optimized for evaluating high-frequency processing are needed to adequately study its effects on speech identification. The goal of the current research was to develop a medium for evaluating the effects of high-frequency processing in hearing aids.
A list of 115 consonant-vowel-consonant-vowel-consonant nonsense syllables with American English consonants in all word positions was created in an open-set phoneme identification format. The source material was spoken by a male and a female speaker. A custom computer program was developed for administration of the test and automatic analysis of the test results. Nine normal-hearing listeners were employed in the collection of the normative data. The test was presented to the listeners in quiet (at 68 dB SPL), in noise at five signal-to-noise ratios (SNRs; -10, -5, 0, 5, and 10), and in a low-pass filter condition with cutoff frequencies at 500, 1000, 1500, 2000, and 4000 Hz. The data were examined to evaluate the psychometric properties of the test for different phoneme positions and phoneme classes. In addition, a shortened version of the test was developed based on the data from normal-hearing listeners. The test-retest reliability was verified at 0 dB SNR. The full and shortened versions of the test were repeated in 10 hearing-impaired listeners at their most comfortable listening level in quiet and in noise at various SNRs.
The availability of high-frequency output was verified with acoustic analysis. The performance intensity functions for both versions of the test (i.e., male and female speakers) showed expected monotonic growth with SNR and cutoff frequencies. High reliability was seen between test and retest identification scores in normal-hearing and hearing-impaired listeners.
The current nonsense syllable test provided a reliable and efficient means for phoneme identification testing.
助听器中的许多新处理功能主要影响高频信息。需要优化用于评估高频处理效果的语音感知测试，以充分研究其对语音识别的影响。本研究的目的是开发一种评估助听器高频处理效果的方法。
创建了一个包含美式英语所有单词位置辅音的 115 个辅音-元音-辅音-元音-辅音无意义音节的列表，采用开放式音位识别格式。源材料由一名男性和一名女性演讲者朗读。开发了一个定制的计算机程序来管理测试和自动分析测试结果。九名正常听力听众参与了规范数据的收集。在安静环境下（68dB SPL）、五个信噪比（SNR；-10、-5、0、5 和 10）下的噪声中以及截止频率为 500、1000、1500、2000 和 4000Hz 的低通滤波器条件下向听众呈现测试。检查数据以评估不同音位位置和音位类别的测试心理测量特性。此外，根据正常听力听众的数据开发了测试的缩短版本。在 0dB SNR 下验证了测试的重测信度。在安静环境中和不同 SNR 下的噪声中，在最舒适的听力水平下，对 10 名听力障碍者重复进行了完整和缩短版本的测试。
通过声学分析验证了高频输出的可用性。两种版本测试（即男性和女性扬声器）的性能强度函数均显示出与 SNR 和截止频率的预期单调增长。在正常听力和听力障碍者中，测试和重测识别分数之间的相关性很高。
目前的无意义音节测试为音位识别测试提供了一种可靠且高效的方法。
