Determination of boundary conditions by inverse modeling of room acoustics

房间声学模拟算法和可听化工具基于关于房间几何形状、反射和散射特性以及诸如衍射和模式的波效应的若干简化。以前的研究和正在进行的研究表明，如果遵循某些程序，这些简化可以被认为是听不见的。然而，所有这些准则都取决于实施、所选择的房间示例以及用于可听化的信号。除了基于物理的模拟之外，基于同时掩蔽的感知剔除允许控制可听化策略，以将性能保持在实时处理的极限内。 目前还不清楚模拟参数如何影响室内声学在其各个维度（响度，音色，宽敞度，清晰度等）的感知。在这方面的模拟参数是：CAD房间模型中的细节程度、吸收和散射模型中的不确定性、在时间和空间分辨率中从早期精确镜面反射模型到统计反射模型的过渡顺序、后期混响（辐射度、漫射雨、成形噪声）及其统计参数（颗粒数量、斑块大小等）的实现（“方言”），申请人先前的工作集中在混合几何建模、更好的表面建模、混合几何-波建模和循环测试。实际的研究项目主要集中在交互式VR系统的实时实施和研究以及大型房间的CAD建模。本研究单元的方法是以前工作的逻辑延续。先前的工作还包括混响滤波器设计中的参数的初步研究。这些研究表明，基于AFC的听力测试策略是非常合格的，通过心理测量功能来测量模拟伪影的可听度阈值。尽管如此，测试条件（仅音频、音频和视觉、音频和视觉和任务）以及心理声学测试在其再现质量以及测试主题的选择和指导方面是该项目的关键部分。 这个子项目的目标是确定感知检测能力和可区分性，相对于模拟参数及其在空间和时间域中的限制。申请人已经开发的现有算法应当应用于心理声学批准的方法并进行测试，以确定关于房间声学感知的各种维度的感知阈值和环境模拟参数的影响。

Room acoustic simulation algorithms and auralization tools are based on several simplifications concerning the room geometry, the reflection and scattering properties and the wave effects such as diffraction and modes. Previous studies and ongoing research have shown that these simplifications can be considered inaudible, if certain procedures are followed. All those guidelines, however, depend on the Implementation, on the room example chosen and on the signal used for auralization. Apart from physically-based simulation, perceptual culling based on simultaneous masking allows control of auralization strategies for keeping the performance in the limits of real-time processing. It is not yet known quantitatively, how the simulation parameters influence the perception of room acoustics In its various dimensions (loudness, timbre, spaciousness, clarity, etc.). Simulation parameters in this respect are: degree of details in the CAD room model, the uncertainty in absorption and scattering models, the transition order from early exact specular to statistic reflection models in temporal and spatial resolution, the implementation ("dialect") of late reverberation (radiosity, diffuse rain, shaped noise) and its statistical parameters (number of particles, patch sizes, etc.), the role of uncertainty of HRTF databases and source recording techniques.Previous work of the applicant was focused on hybrid geometric modeling, better surface modelling, hybrid geometric-wave modeling and round robin tests. Actual research projects are focused on real-time Implementation and studies of interactive VR systems and on CAD modelling of large rooms. The approach In this Research Unit is the logical continuation of previous work. Previous work also included preliminary studies of the parameters In reverberation filter design These studies showed that AFC based listening test strategies are well qualified to measure audibility thresholds of simulation artifacts by means of psychometric functions. Nevertheless the test conditions (only audio, audio and vision, audio and vision and task) and the psychoacoustic test in its reproduction quality and test subject selection and instruction Is a crucial part of the project. The goal of this subproject Is to determine the perceptual detection ability and differentiability, with respect to the simulation parameters and their limits in spatial and temporal domains. Existing algorithms that have been developed by the applicant should be applied and tested with psycho-acoustically approved methods, to ascertain perceptual thresholds regarding various dimensions of room acoustics perception and the impact of environmental simulation parameters.