Speech Perception Enhancement Using Novel Signal Processing in Bimodal Hearing

项目摘要/摘要 人工耳蜗(CI)与助听器(HAS)联合使用对言语感知的影响 耳朵(即双峰听力)差异很大。这种可变性取决于用户处理频率和 对言语感知至关重要的时间信息。通过识别和增强这些声学信息，语音 观感将显著改善。在这一领域的项目中，我们的目标是建立和核实一个量身定做的身份 负责辅音识别的光谱和时间提示的方案。我们最近的双峰研究 结果表明，辅音的某些频率范围和时间段对辅音增强至关重要 (称为目标频率或时间范围)，而其他频率和时间范围会导致辅音混淆 (称为“冲突的频率或时间范围”)。我们的清晰度指数-Gram(AI-Gram)信号处理可以添加 并压制这些目标和冲突射程上的强度。在目标1中，我们将确定死者的影响 辅音识别的区域。然后将在单个对象上识别目标范围和冲突范围 每个辅音的基础是单独的HA、单独的CI和静音的CI+HA。目标频率范围将为 通过找到产生显著辅音增强的频率区域来确定，而冲突的 频率范围将通过找到造成辅音混淆的频率区域来确定。目标是 时间范围将通过查找负责戏剧性辅音的辅音片段来确定 在系统地截断辅音的同时进行改进。目标时间范围将用作冲突 时间范围，因为冲突的频率范围将是影响目标的最有害因素 频率范围，如果它们在时间上重合。在目标2中，辅音识别将在安静和噪声中进行测量 在三种AI-Gram处理条件下：1)目标范围单独，+6分贝；2)冲突范围 单独抑制-6分贝；以及3)增强目标和抑制冲突距离。对于每个人工智能- 在语法处理条件下，辅音识别将在匹配的收听条件中测量(例如， 仅在HA中识别的目标或冲突范围将仅在HA监听条件中呈现)。 为了确定单边检测能力如何影响双峰效应，在目标上处理辅音 或者仅在HA和CI中标识的冲突范围将各自呈现给CI+HA监听 条件。这项拟议的工作将识别有助于双峰受益的声学提示，并将揭示 这些线索在不同的通道中被整合或干扰。确定目标和目标的相对影响 单独在HA、CI和CI+HA中对AI-Gram敏感辅音的冲突范围 将有助于确定HA和CI的上限和下限截止频率，并微调这些截止频率 频率。这些数据对于长期目标来说是非常需要的：开发一种量身定做的双峰拟合程序。 该地区项目每年将为贝勒大学的四名本科生提供临床研究机会。

Project Summary/Abstract Speech perception for those who use cochlear implants (CIs) in combination with hearing aids (HAs) in opposite ears (i.e., bimodal hearing) varies greatly. This variability depends on the users’ ability to process frequency and time information critical for speech perception. By identifying and enhancing this acoustic information, speech perception will significantly improve. In this AREA project, we aim to establish and verify a tailored identification scheme for the spectral and temporal cues responsible for consonant recognition. Our recent bimodal study shows that some frequency ranges and time segments of consonants are critical for consonant enhancement (called “target frequency or time ranges”) while other frequency and time ranges cause consonant confusions (called “conflicting frequency or time ranges”). Our Articulation Index-Gram (AI-Gram) signal processing can add and suppress intensity on these target and conflicting ranges. In Aim 1, we will determine the effect of the dead regions on consonant recognition. Target and conflicting ranges will then be identified on an individual subject basis for each consonant in the HA alone, CI alone, and CI+HA in quiet. The target frequency range will be determined by finding the frequency regions creating dramatic consonant enhancement, while the conflicting frequency ranges will be determined by finding the frequency regions creating consonant confusion. The target time ranges will be determined by finding the segment of the consonants responsible for dramatic consonant improvement while systematically truncating the consonant. The target time range will be used as the conflicting time ranges because the conflicting frequency ranges would be the most detrimental factor affecting the target frequency ranges if they coincide in time. In Aim 2, consonant recognition will be measured in quiet and noise under the three AI-Gram processing conditions: 1) target ranges alone with +6 dB gain; 2) conflicting ranges alone with -6 dB suppression; and 3) both intensified target and suppressed conflicting ranges. For each AI- Gram processing condition, consonant recognition will be measured in the matched listening conditions (e.g., the target or conflicting ranges identified in the HA alone will be presented in the HA alone listening condition). To determine how the unilateral detection ability affects bimodal benefit, the consonants processed on the target or conflicting ranges identified in the HA alone and CI alone will each be presented to the CI+HA listening condition. This proposed work will identify acoustic cues that contribute to bimodal benefit and will reveal how these cues are integrated or interfered with across modalities. Defining the relative impact of the target and conflicting ranges on the AI-Gram-sensitive consonants in the HA alone, the CI alone, and the CI+HA together will help determine the upper and lower cutoff frequencies of a HA and a CI and fine-tune these cutoff frequencies. This data is much needed for the long-term goal: developing a tailored bimodal fitting procedure. The AREA project will provide clinical research opportunities for four undergraduate students per year at Baylor.