Auditory Neural Coding of Speech

描述（由申请人提供）：大多数语音对话发生在房间表面的竞争声音和声学反射的存在下。听力受损的人经常抱怨在如此复杂的声音环境中理解语言有困难，即使他们在安静的环境中表现良好。在这里，我们提出了神经生理学和计算研究，以解决复杂环境中听力的两个方面的基本问题：(1)听觉系统如何补偿混响引起的声信号退化；(2)听觉系统如何提取谐波复音的音高，这是听者用来区分同时声源的主要线索之一。我们将记录听神经（AN）、耳蜗腹侧核（VCN）和下丘（IC）的单个单元对复杂声音的反应，这些声音包含了复杂声环境的一些特征，并开发预测这些反应的计算模型。具体目标1是测试听觉系统包含神经机制的假设，使其能够在混响中保持良好的方向和时间敏感性。我们将测量IC神经元在模拟室内环境中的方向和时间包络灵敏度，将这些响应与现有双耳处理模型的预测结果进行比较，并开发包含混响补偿机制的新模型。目的2是验证一个假设，即耳蜗行波对复杂音调的音高产生强大的时空线索，这些线索可以通过神经机制提取，神经机制对调谐到略微不同频率的AN纤维的spike放电的相对时间敏感。我们将测试这些时空音高线索在神经网络中的可用性和鲁棒性，然后检查这些线索是否可以被已知对单声相位敏感的VCN神经元提取。本研究探讨了听觉理论中的一些基本问题，如音高处理的神经机制、混响中回声抑制的机制和声源分离的机制。这可能会让我们更好地理解为什么听力受损和老年听众在混响和竞争声音的存在下比正常听众更难理解讲话，并可能有助于开发新型助听器和听觉（耳蜗和脑干）植入物，这些植入物在具有挑战性的环境中表现更好。

DESCRIPTION (provided by applicant): Most speech conversations occur in the presence of competing sounds and acoustic reflections from room surfaces. Hearing-impaired people often complain of difficulties understanding speech in such complex acoustic environments even if they do well in quiet. Here we propose neurophysiological and computational studies that address fundamental questions about two aspects of listening in complex environments: (1) How the auditory system compensates for the degradation in the acoustic signal caused by reverberation; (2) How the auditory system extracts the pitch of harmonic complex tones, one of the main cues used by listeners to segregate simultaneous sound sources. We will record from single units in the auditory nerve (AN), ventral cochlear nucleus (VCN) and inferior colliculus (IC) in response to complex sounds that incorporate some features of complex acoustic environments, and develop computational models that predict these responses. Specific Aim 1 is to test the hypothesis that the auditory system contains neural mechanisms that allow it to preserve good directional and temporal sensitivity in reverberation. We will measure the directional and temporal envelope sensitivity of IC neurons in simulated room environments, compare these responses with predictions of existing models of binaural processing, and develop new models incorporating reverberation compensation mechanisms. Aim 2 is to test the hypothesis that the cochlear traveling wave creates robust spatio-temporal cues to the pitch of complex tones that can be extracted by a neural mechanism sensitive to the relative timing of spike discharges from AN fibers tuned to slightly different frequencies. We will test the availability and robustness of these spatio-temporal pitch cues in the AN, then examine whether these cues can be extracted by neurons in the VCN known to be sensitive to monaural phase. This research addresses fundamental issues in auditory theory such as the neural mechanisms for pitch processing, the mechanisms for echo suppression in reverberation, and mechanisms of sound source segregation. It may lead to a better understanding of why hearing-impaired and elderly listeners have greater difficulties understanding speech in the presence of reverberation and competing sounds than do normal listeners, and may help develop new kinds of hearing aids and auditory (cochlear and brainstem) implants that perform better in challenging environments.