Neural Correlates of Streaming of Complex Sounds

项目摘要 这项研究的目的是解释听觉系统是如何分析复杂的 声学场景，其中多个声音“流”(例如，并发声音)相互作用 争夺人们的注意力。协同是通过将同时的行为结合在一起而产生的 以及雪貂初级和次级听觉皮质的神经生理学测量 2)具有可比的行为和脑电(EEG)测量 3)时间的理论和计算框架中的人类 连贯假说。具体目标1涉及单个单位皮层的记录 当雪貂分离语音混合并检测目标的存在时，响应 目标说话者所说的话。这些实验将与人类试验平行进行。 由浊音(和声)的自然混合组成的语音流 和无声(类似噪音)的声音。具体目标2探讨了连贯性和 注意使用具有更简单频谱的刺激进行流绑定和分离 特征，例如纯音和噪声序列。我们的目标是测试强大 时间相干性假说的预测 时间同步的、交替的或重叠的声音序列。具体目标3 将此方法扩展到更复杂刺激的更高级别属性，如音调 以及音色，它们在生态上更具相关性。在这里，我们研究了音调和 音色刻画流的连续性，并将 在雪貂和人类中都有。这项研究具有直接和重要的健康意义。 暗示，因为听力受损的人最常见的抱怨之一 个人(包括助听器佩戴者或人工耳蜗植入者)是他们发现的 很难分离并发的声音流，也很难有选择地注意以下其中之一 这些流(如某人的声音)在其他流中。一个更清晰的 理解感知分离能力背后的机制，以及 注意，听觉流可能会使我们更清楚地了解 这些选择性听力的困难，可能会启发更有效的设计 用于助听器、人工耳蜗和自动助听器的声音分离算法 语音识别设备。

Project Summary The goal of this research is to explain how the auditory system analyzes complex acoustic scenes, in which multiple sound “streams” (e.g., concurrent voices) interact and compete for attention. Synergy is created by combining 1) simultaneous behavioral and neurophysiological measures from primary and secondary auditory cortex in ferrets with 2) comparable behavioral and electroencephalographic (EEG) measures in humans within 3) the theoretical and computational framework of the temporal coherence hypothesis. Specific Aim 1 involves recordings of single-unit cortical responses while ferrets segregate speech mixtures and detect the presence of a target word spoken by the target talker. The experiments will be paralleled by tests of human streaming of speech sounds that are comprised of natural mixtures of voiced (harmonic) and unvoiced (noise-like) sounds. Specific Aim 2 explores the role of coherence and attention in stream binding and segregation using stimuli with simpler spectral characteristics, e.g., pure-tones and noise sequences. The goal is to test strong predictions of the temporal coherence hypothesis with respect to the effects of temporally synchronous, alternating, or overlapping sound sequences. Specific Aim 3 extends this approach to higher-level attributes of more complex stimuli, such as pitch and timbre, which are more ecologically relevant. Here we examine the role of pitch and timbre in characterizing the continuity of a stream, and binding the elements of the stream in both ferrets and humans. This research has direct and significant health implications, because one of the most common complaints of hearing-impaired individuals (including wearers of hearing aids or cochlear implants) is that they find it difficult to separate concurrent streams of sounds, and to attend selectively to one of these streams (such as someone's voice) among other streams. A clearer understanding of the mechanisms underlying the perceptual ability to separate, and attend to, auditory streams will likely lead to a clearer understanding of the origin of these selective-listening difficulties, and it may inspire the design of more effective sound-separation algorithms for use in hearing aids, cochlear implants, and automatic speech recognition devices.