Acoustic and perceptual effects of WDRC amplification

WDRC 放大的声学和感知效果

基本信息

批准号：
8303019
负责人：
PAMELA E. SOUZA
金额：
$ 22.45万
依托单位：
NORTHWESTERN UNIVERSITY
依托单位国家：
美国
项目类别：
财政年份：
2005
资助国家：
美国
起止时间：
2005-03-01 至 2015-07-31
项目状态：
已结题

项目摘要

DESCRIPTION (provided by applicant): An ongoing goal of our work has been to understand the effects of wide dynamic range compression (WDRC) amplification on temporal cues for speech recognition. WDRC is the default configuration in all digital aids, so the extent to which WDRC improves speech recognition dictates hearing aid benefit and use. However, WDRC can degrade the modulation properties of the signal, particularly the low-modulation rate (envelope) cues which are important for speech recognition. This project encompasses three sets of experiments that take a systematic approach to prescribing amplification parameters. The first specific aim is to define the extent to which envelope cues can be altered without degrading speech recognition. We will quantify how much envelope cues can be altered, particularly under conditions of spectral degradation similar to those experienced by listeners with hearing loss. We are most concerned with listeners with severe hearing loss who depend more on envelope cues, and who have received little attention in the literature. The second specific aim is to determine whether the acceptable level of envelope distortion depends on factors associated with degree of hearing impairment. The working hypothesis is that listeners who have reduced access to spectral cues, and therefore must rely more on envelope information, will have lower recognition scores due to the envelope alteration caused by WDRC. We will use several measures of ability to detect and use spectral differences necessary for speech recognition, which will be related to the speech-recognition changes due to envelope distortion. The third specific aim is to determine whether the acceptable amount of envelope distortion depends on the listening environment. We will consider the effect of two factors on the acceptable level of envelope distortion: effect of background noise and length of speech utterance. Ultimately, the goal of this work is to determine the range of compression values that maximize the restoration of audibility without distorting usable modulation cues. PUBLIC HEALTH RELEVANCE: Hearing loss is one of our most pervasive health problems but the only treatment option for most of those with hearing loss is a hearing aid. To improve hearing aid satisfaction and benefit and avoid distorting important speech cues, we need to choose appropriate processing parameters for each individual. The focus of this project is to determine hearing aid settings that will provide the best possible speech recognition for each individual under a variety of listening conditions.

描述（由申请人提供）：我们工作的一个持续目标是了解广泛的动态范围压缩（WDRC）放大对语音识别的时间提示的影响。 WDRC是所有数字辅助工具中的默认配置，因此WDRC改善语音识别的程度决定了助听器的利益和使用。但是，WDRC可以降低信号的调制特性，尤其是对于语音识别很重要的低调节率（包膜）线索。该项目包括三组实验，这些实验采用了一种系统的方法来规定放大参数。第一个具体目的是确定可以在不降低语音识别的情况下改变信封提示的程度。我们将量化可以更改多少信封提示，尤其是在光谱降解条件下，类似于听力损失的听众所经历的条件。我们最关心的是听力损失严重的听众，他们更多地依赖信封提示，并且在文献中很少关注。第二个具体目的是确定可接受的信封失真水平是否取决于与听力障碍程度相关的因素。工作假设是，由于WDRC引起的信封变化，降低了访问光谱线索并必须更多地依靠信封信息的听众将具有较低的识别分数。我们将使用几种能力来检测和使用语音识别所需的光谱差异，这将与由于包膜失真引起的语音识别变化有关。第三个具体目的是确定可接受的信封失真数量是否取决于听力环境。我们将考虑两个因素对信封失真的可接受水平的影响：背景噪声和语音话语的影响。最终，这项工作的目的是确定压缩值的范围，以最大程度地恢复了听觉性，而不会扭曲可用的调制线索。公共卫生相关性：听力损失是我们最普遍的健康问题之一，但是对于大多数听力损失的人来说，唯一的治疗方法是助听器。为了提高助听器满意度并受益，并避免扭曲重要的语音提示，我们需要为每个人选择适当的处理参数。该项目的重点是确定助听器设置，这些设置将在各种听力条件下为每个人提供最佳的语音识别。