Real-time auralization of time-varying sound fields with moving sound sources and listeners

本研究针对具有快速移动声源和听众的声学虚拟场景进行实时可听化。近年来，开发了先进的方法，允许实时基于物理的声学模拟（几何声学，衍射模型，声音传输模型等）。移动源和听众和由此产生的多普勒效应已被模拟使用简化的方法。在声源和听众之间的距离较大的情况下，这些方法无法以合理的方式模拟声音传播并且没有伪影。移动物体的轨迹在运行时只能在有限的程度上改变，这对虚拟环境中的交互产生了很大的限制。基于我们在这一科学领域的长期经验，本项目的目的是增强最近的方法，并开发新的快速移动源和听众的实时仿真方法。我们的目标不仅是实时执行可听化，在虚拟场景中实现被动的听觉体验，而且特别关注用户交互，这对虚拟现实至关重要。场景中所有对象的运动（轨迹）都应该是用户可以实时修改的。为了实现这一目标，在以下领域需要新的科学发现：a）实时运动建模，B）实时地物理正确地计算声音传播时间，以及c）考虑可听化中的时间正确的参数，这要求描述虚拟场景中的变量沿着时间历史。然后，这些结果将被组合成一个统一的听觉化方法。这不仅需要了解这些单独的领域，而且还需要了解它们之间的相互依赖关系（运动建模，声音传播，场景数据结构）。所获得的科学结果将被应用到现有的RWTH亚琛实时可听化系统中，使其能够作为未来增强研究的工具（例如，城市环境中的综合噪声场景）。本项目考虑的潜在声源为汽车、高速列车和飞机。特别是对于噪音研究，如交通和飞机规划，这种虚拟环境可以证明是一个有益的工具，在未来。最初的两年资助期需要开发所有必要的基础和软件工具的实施。随后的第三年致力于主观用户研究与统计心理声学评估。在此，应详细比较潜在的方法和参数，以便针对应用进行优化。

This research proposal deals with the real-time auralization of acoustic virtual scenes with fast moving sound sources and listeners. In recent years, sophisticated methods were developed, which allow the physically-based simulation of acoustics in real-time (geometrical acoustics, diffraction models, sound transmission models, etc.). Moving sources and listeners and the resulting Doppler effects have been simulated using simplified approaches only. In case of large distances between sources and listeners, these methods fail to simulate the sound propagation in a plausible way and free of artifacts. Trajectories of moving objects can be altered during runtime only to a limited degree, which puts a strong limitation on the interaction in virtual environments.Based on our long-term experience in this scientific domain, the aim of this project is to enhance recent approaches and to develop new real-time simulation methods for fast moving sources and listeners. The objective is not only to perform the auralization in real-time, enabling a passive listening experience in the virtual scene, but especially to focus on the user interaction, which is essential for Virtual Reality. Movements (trajectories) of all objects in the scene should be modifiable by the user in real-time. In order to achieve this goal new scientific findings are needed in the following areas: a) real-time motion modeling, b) physically-correct computation of sound propagation times in real-time and c) consideration of the temporally-correct parameters in the auralization, which demands the description of the variables in a virtual scene along a temporal history. These results will then be combined into a unified auralization approach. This requires not only an understanding of these individual fields, but also of their distinct interdependencies (motion modeling, sound propagation, scene data structures).The obtained scientific results will be implemented into the existing RWTH Aachen real-time auralization system, enabling it to serve as a tool for enhanced studies in the future (e.g. comprehensive noise scenarios in urban surroundings). The potential sound sources regarded in this project are cars, high-speed trains and aircrafts. Especially for noise research, like traffic and aircraft planning, such virtual environments could prove as a beneficial tool in the future.The initial two-year funding period is needed for the development of all required foundations and the implementation of the software tools. A subsequent third year is dedicated to subjective user studies with statistical psychoacoustic assessments. Here, the potential methods and parameters shall be compared in detail in order to optimize them for the application.