Perception of Dynamic Acoustic Signals by Hearing Impaired Listeners

描述（由申请人提供）：语音是一种内在的动态信号，频谱动态是辅音和元音感知不可或缺的一部分。形成峰运动的速度、方向和程度为辅音和元音的同一性提供了线索。与听力损失和衰老相关的外周听觉敏感性、精确的时间处理和频率选择性的丧失将破坏对峰运动的感知。因此，减少或消除动态频谱线索可能会损害语音理解，特别是在不利的听力条件下，如背景噪声或混响。本研究项目的目的是了解听障听众对动态声学信号的感知，长期目标是提高这些人的言语理解，特别是在不利的听力条件下。这项拟议的研究比较了听力正常的听众和听力受损的听众在一些任务上的表现，这些任务旨在评估他们对复杂的、频率随时间变化的非言语听觉刺激的基本感知能力。实验程序将包括频谱变化检测和识别的自适应测试，这些测试在非语音刺激中具有与语音刺激相同的时间和频率特征，以及在几种竞争背景噪声和模拟混响中估计频谱变化检测和识别阈值。语音识别能力在背景噪声和混响也将进行评估。基于这些频谱变化感知和语音识别的行为测量结果，将开发一个计算模型，该模型可以解释由于听力状态和刺激特征导致的语音可理解性差异。计算模型的有效性将通过比较模型预测和实际听者的表现来评估。将频谱变化感知信息应用到语音识别模型中，将对开发新的助听器处理策略和康复听觉训练计划产生影响，这些计划可能会提高听障人士的语音清晰度，减少听障人士的听力努力。

DESCRIPTION (provided by applicant): Speech is an inherently dynamic signal and spectral dynamics are integral to perception of both consonants and vowels. The rate, direction, and extent of formant movements provide cues to consonant and vowel identity. The loss of peripheral auditory sensitivity, precise temporal processing, and frequency selectivity associated with hearing loss and aging will disrupt the perception of formant movements. As a result, the reduction or elimination of dynamic spectral cues may impair speech understanding, especially in adverse listening conditions such as background noise or reverberation. The purpose of this program of research is to understand the perception of dynamic acoustic signals by hearing-impaired listeners, with the long-term goal of improving speech understanding by these individuals, especially in adverse listening conditions. The proposed research compares the performance of normally-hearing and hearing-impaired listeners on a number of tasks designed to assess basic perceptual abilities in response to complex, non-speech auditory stimuli that change in frequency over time. Experimental procedures will consist of adaptive tests of spectral change detection and discrimination in non-speech stimuli which possess timing and frequency characteristics to those of speech, and estimation of spectral change detection and discrimination thresholds in several types of competing background noise and with simulated reverberation. Speech recognition ability in background noise and reverberation will also be assessed. Based on the results of these behavioral measures of spectral change perception and speech recognition, a computational model will be developed that can account for differences in speech intelligibility due to hearing status and stimulus characteristics. The effectiveness of the computational model will be evaluated by comparing model predictions with actual listener performance. The application of information regarding spectral change perception to a model of speech recognition will have implications for the development of new hearing aid processing strategies and rehabilitative auditory training programs that may enhance speech intelligibility and reduce listening effort in hearing-impaired individuals.