An Encepholographic Investigation of Spatial Cue Augmentation

人类听觉系统的空间定位能力是听觉感知的一个基本方面。这种能力不仅有助于在三维空间中定位声源，还可以在复杂的声音场景中选择性地关注特定的声源，这种现象通常被称为“鸡尾酒会效应”。其中的关键是用于水平定位的声间时间和电平差异（IT和ILD），以及用于高程和模糊源位置的单声光谱线索。这些线索是高度个性化的，完全由每个听者独特的形态决定。因此，非个性化的线索会对空间感知产生不利影响。不能保存这些线索的助听器会损害听障（HI）个体的空间感知能力。虽然双耳线索保存可以提高语音清晰度和定位，但HI听众可能仍然面临由于频率特异性听力损失的挑战。对非个体空间线索的适应已经被证明，Armstrong等人的研究结果表明，个体HRTF测量可能并非对空间感知的所有方面都是最好的，这表明需要针对特定应用选择HRTF。这些发现不仅强调了定制空间线索的重要性，而且提出了空间线索是否可以增强空间感知的特定行为方面的问题。空间听力评估传统上依赖于两种主要方法：频谱失真和更先进的听觉模型等数值测量方法，以及定位测试等主观听力测试。然而，这两种方法都有缺点。数字指标可能会错过感知上的细微差别，而主观测试可能会受到个人识别能力的影响。因此，需要对空间感知进行更客观的测量。脑电图（EEG）提供了希望。脑电图是一种记录大脑活动的非侵入性方法，可以深入了解听觉感知的潜在神经机制和途径。EEG可以检测到事件相关电位（event- correlation potential, ERPs），这在听觉空间感知的背景下，可以为大脑如何处理和整合空间线索的时间动态提供有价值的见解。机器学习（ML）的进步增强了对脑电图数据的解释，有可能为助听器和消费电子产品等设备的空间感知动态评估提供信息。相关空间线索的重要性、听者对非个体线索的适应能力、空间线索增强的前景以及基于脑电图的空间听力效果评估方法激发了本博士的研究目标，包括：(1)探索空间线索改变对空间听力和言语清晰度的行为影响；(2)设计增强空间线索的方法以增强空间感知的特定方面；(3)评估空间线索增强的神经关联。

The human auditory system's spatial localisation capability is a fundamental aspect of our auditory perception. This ability not only facilitates the localisation of sound sources in three-dimensional space but also enables selective attention to specific sound sources in complex sound scenes, a phenomenon popularly termed the 'cocktail party effect'. Key to this are interaural time and level differences (IT and ILD) for horizontal localization, and monaural spectral cues for elevation and ambiguous source locations.These cues are highly individual, determined solely by each listener's unique morphology. Consequently, non-individualised cues can adversely impact spatial perception. Hearing aids that fail to preserve these cues can impair spatial perception in hearing-impaired (HI) individuals. While binaural cue preservation can enhance speech intelligibility and localization, HI listeners may still face challenges due to frequency-specific hearing loss. Adaptation to non-individual spatial cues has been shown, and Armstrong et al.'s findings suggest that individual HRTF measurements may not be the best for all aspects of spatial perception, indicating the need for application-specific HRTF selection. These findings not only underscore the importance of tailored spatial cues but also pose the question of whether spatial cues can be augmented to enhance specific behavioural aspects of spatial perception.Spatial hearing assessment traditionally relies on two main methods: numerical measures like spectral distortion and more advanced auditory models, as well as subjective listening tests such as localisation tests. However, both approaches have drawbacks. Numerical metrics may miss perceptual nuances, and subjective tests can be skewed by individual discernment abilities. Thus, there is a need for more objective measures of spatial perception. Electroencephalography (EEG) offers promise. EEG is a non-invasive method for recording brain activity, providing insights into the underlying neural mechanisms and pathways of auditory perception. EEG can detect event-related potentials (ERPs), which in the context of auditory spatial perception, canprovide valuable insights into the temporal dynamics of how the brain processes and integrates spatial cues. Advances in machine learning (ML) enhance the interpretation of EEG data, potentially informing the dynamic assessment of spatial perception in devices like hearing aids and consumer electronics.The importance of pertinent spatial cues, listener ability to adapt to non-individual cues, the promise of spatial cue augmentation and EEG-based evaluation methods of spatial hearing efficacy motivate the aims of this PhD comprising (1) Exploring the behavioural impact of altered spatial cues on spatial hearing and speech intelligibility, (2) Designing methods to augment spatial cues to enhance specific aspects of spatial perception and (3) Evaluating the neural correlates of spatial cue augmentation.