Auditory perception of sound reflections and source localization in dynamic scenes

除了视觉信息外，声音信息使人类能够在周围环境中定位自己。在烟雾弥漫的房间、黑暗或雾蒙蒙的环境或视障人士的紧急情况下尤其如此。在复杂的环境中，人类利用早期的边界反射（墙壁、天花板、地板、大型物体）及其与声源的关系来确定声学方向。本项目将研究复杂声学场景的听觉感知。不同的场景，如自由的田野场景，大房间，或户外场景（如在一个城市的方向）将被评估。目前，在实时室内声学模拟中，利用图像源产生可感知的早期反射。如果墙壁的相对方向或位置由于环境几何模型的近似而发生变化，那么图像源的位置也会发生变化。这样，对这个场景的听觉感知就会发生变化。为这个项目提供基础的环境模型将只包含主导的早期反思。这意味着在非常大的房间里，外部空间和靠近墙壁的听者位置没有被考虑在内。首先，将使用简单的几何房间模型分析听力正常（非视障）的人的感知。边界界面的最大不可听变化（距离/旋转）取决于双耳参数、耳间时间和水平差以及心理声学因素。在第一阶段，应研究双耳参数、心理声学因素和（少数）图像源与墙壁尺寸、头部运动和受试者与墙壁距离的关系。在第二阶段，将考虑头部运动以及源和听者之间的距离，研究虚拟声环境的场景感知。几个不同位置的光源将被整合到测试对象的场景中。相干度，即信号的相似性，应降低到最低限度，以避免虚影源。通过实验确定声源与被试之间的距离以及头部运动对最大可听声源数的影响。该项目的成果可用于在虚拟声学现实中创建房间模型。另一个可能的应用是视障人士的虚拟听觉训练，使他们能够在城市或未知的环境中确定自己的方向。

Apart from visual information, acoustic information enables human beings to orient themselves in their surroundings. This is especially true for emergencies in smoke-filled rooms, dark or foggy environments, or in case of visually impaired people. In complex environments, human beings use early boundary reflections (walls, ceilings, floors, large objects) and their relation to a sound source for an acoustic orientation. The auditory perception of complex acoustic scenes will be studied for this project. Different scenes such as free field scenes, large rooms, or outdoor scenes (e.g. orientation in a city) will be evaluated. Nowadays image sources are used to generate the perceived early reflections for real-time room acoustic simulations. If the relative orientation or the position of a wall changes due to an approximation of the geometric model of the environment, the position of the image source will change as well. Thus, the auditory perception of this scene will change. The environmental model that provides the basis for this project will only contain dominate early reflections. This means that exterior spaces and listener positions that are close to the walls in very large rooms are not taken into account. First of all, the perception of people with normal hearing (not visually impaired) will be analyzed using simple geometric room models. The maximum non-audible variation of a boundary interface (distance/rotation) depends on binaural parameters, interaural time and level difference, and psycho-acoustic factors. In the first phase, the correlation of binaural parameters, psychoacoustic factors, and the (few) image sources shall be studied as a function of the wall size, the head movement, and the distance of the test subject from the wall. In the second phase, the scene perception of a virtual acoustic environment will be studied taking into account the head movement as well as the distance between the source and the listener. Several sources with different positions will be integrated into the scenes for the test subjects. The degree of coherence, i.e. the similarity of the signals, shall be reduced to a minimum, to avoid phantom sources. Experiments will be carried out to determine the influence of the distance between sound source and test subject, and the head movement on the maximum number of audible sources. The results of this project can be used to create a room model in the virtual acoustic reality. Another possible application is the virtual, acoustic training of visually impaired persons to enable them to orient themselves in a city or in an unknown environment.