Leijon A, Lindkvist A, Ringdahl A, Israelsson B
Department of Information Theory, Chalmers University of Technology, Gothenburg, Sweden.
Ear Hear. 1991 Aug;12(4):251-60. doi: 10.1097/00003446-199108000-00004.
Four different prescriptions of hearing aid insertion gain versus frequency were validated with a group of 26 moderately hearing-impaired, elderly hearing aid users. Three prescriptions were based on calculating the loudness and articulation index (AI) for aided speech, and ranged from a frequency response with moderate high-frequency emphasis, restoring normal loudness for speech peaks, to a response with the greatest high-frequency emphasis, maximizing the AI. The fourth prescription was a well-established formula of the half-gain type. The frequency responses were evaluated by paired comparison ratings of the pleasantness and intelligibility of speech in noise, and by speech identification tests in noise. The subjects rated the flattest response as significantly more pleasant than the other responses, and significantly more intelligible than the prescription with maximal high-frequency emphasis. There were no detectable differences in signal/noise ratios required for 50% speech identification. These results indicate that a prescription that restores normal loudness for speech peaks in each critical band is probably more easily accepted than either a procedure which is intended to make all speech bands equally loud, or a prescription which maximizes the AI.
针对一组26名中度听力受损的老年助听器使用者，验证了四种不同的助听器插入增益与频率的处方。其中三种处方是基于计算助听语音的响度和清晰度指数（AI），其频率响应范围从中度高频强调、恢复语音峰值的正常响度，到最大高频强调、使AI最大化。第四种处方是一种成熟的半增益类型公式。通过对噪声中语音的愉悦度和可懂度进行配对比较评分，以及在噪声中进行语音识别测试来评估频率响应。受试者将最平坦的响应评为比其他响应明显更愉悦，并且比具有最大高频强调的处方明显更易懂。在50%语音识别所需的信噪比方面没有可检测到的差异。这些结果表明，一种在每个临界频带中恢复语音峰值正常响度的处方可能比旨在使所有语音频带响度相等的程序或使AI最大化的处方更容易被接受。
