A multi-modal streaming approach to understanding aided speech perception in noise: discovering relationships between device limitations, perceptual abilities, and neuro-modulation of feature coding

项目总结/摘要 据估计，只有大约七分之一的美国听力损失者使用听力辅助设备。 尽管普遍接受的观点是诸如助听器（HA）的设备改善听力健康和质量 生命采用率低的最常见原因之一是，当试图遵循 在噪音中交谈。为了解决这个问题，大多数高级HA采用自适应定向 处理（DP），其降低被认为不太相关的位置处的增益，从而提高信噪比 （SNRs）更相关的演讲。然而，这种策略容易受到处理延迟和不准确的影响。 语音定位，虽然HA DP已经假定了好处，但与之相关的神经机制并不 很好理解。人们会期望HA通过增加SNR来支持声学特征表示; 然而，如果收听者的语音定位在噪声中较差并且HA DP延迟或不准确，则较差的感知- 特征表示（例如，空间）可能会阻止有效的听觉流。过去的调查评估了 DP相对于无辅助或全向放大的益处，并且已经假设了准确的适应， 使用了固定方向的焦点。此外，受益通常通过自我报告，倾听努力或句子来衡量 噪音中的识别这种传统的方法有重要的局限性：1）研究缺乏系统的控制HA DP，因为中间人放置的障碍，和2）结果只提供了间接的措施的机制 对假定的优势负责。相比之下，在这里，我们将受控HA DP评估与一个强有力的 建立了基于听觉对象的注意理论模型。在这样做的时候，我们可以直接将利益 HA DP与相关的神经机制，我们可以更精确地针对未来HA的创新工作， 技术.我们的多模式流方法包括1）HA DP的系统控制，2） 基本感知能力，以及3）支持听觉流的神经处理的测量。目标1： 在新的噪声中语音任务期间参数地改变增益降低的空间和时间精度， 直接评估焦点和时间转换到自由场中的语音位置的效果（Exp. 1）。在目标2中， 我们使用我们的多模态流方法来测量助听器对空间调谐曲线的影响， 当被动聆听时的清晰度（Exp. 2a）和当出席空间位置（Exp. 2b）。观察到 然后分析DP策略的好处，并确定潜在的心理物理措施 HA DP的预测因子（目的3）。这项研究的结果将提供一个新的理解辅助语音 感知噪声和指导未来HA的发展建立在听觉流理论的基础上。 计划中的R01研究将进一步研究辅助声学和感知特征编码以及听觉 流在一个基于对象的注意力模型。

PROJECT SUMMARY/ABSTRACT It is estimated that only about 1 in 7 Americans who have hearing loss use some hearing assistive device despite the generally accepted view that devices such as hearing aids (HAs) improve hearing health and quality of life. Among the most common reasons for low adoption is an unclear benefit when trying to follow a conversation in the presence of noise. To address this problem, most premium HAs employ adaptive directional processing (DP) that reduces gain at locations considered less relevant, thus improving signal-to-noise ratios (SNRs) of more relevant speech. This strategy, however, is susceptible to processing delays and inaccurate speech localization, and though HA DP has assumed benefits, the neural mechanisms associated with it are not well understood. One would expect that HAs support acoustic-feature representation by increasing SNR; however, if a listener's speech localization is poor in noise and HA DP is delayed or inaccurate, poor perceptual- feature representation (e.g., space) may prevent effective auditory streaming. Past investigations have assessed DP benefits relative to unaided or omnidirectional amplification and have either assumed accurate adaptation or used a fixed directional focus. Moreover, benefit is typically gauged by self-report, listening effort, or sentence recognition in noise. This traditional approach has important limitations: 1) studies lack systematic control of HA DP because of manufacturer-placed barriers, and 2) results provide only indirect measures of the mechanisms responsible for the presumed advantages. In contrast, here we link controlled HA DP evaluation to a strong and established theoretical model of auditory object-based attention. In doing so, we can directly associate benefits of HA DP with relevant neural mechanisms, and we can more precisely target innovation efforts for future HA technology. Our multi-modal streaming approach incorporates 1) systematic control of HA DP, 2) an assay of basic perceptual abilities, and 3) measures of neural processing that support auditory streaming. In Aim 1, we parametrically vary the spatial and temporal precision of gain reduction during a novel speech-in-noise task, to assess directly the effects of focal and timelier transitions to speech location in the free field (Exp. 1). In Aim 2, we use our multi-modal streaming approach to measure the effect of hearing aids on spatial tuning curve sharpness both when passively listening (Exp. 2a) and when attending to a spatial location (Exp. 2b). Observed benefits in DP strategies are then analyzed with respect to psychophysical measures identified as potential predictors of HA DP (Aim 3). The results of this study will deliver a novel understanding of aided speech perception in noise and guide future HA development grounded in an established theory of auditory streaming. Planned R01 research will futher investigate aided acoustic and perceptual feature coding and auditory streaming in an auditory-object based model of attention.