Auralization of Urban Environments: Benchmark and Evaluation

通过该研究项目，我们希望为城市环境的可听化创造更广泛的基础。虚拟声环境可听化的挑战对传播仿真和数字信号处理的方法和实现提出了很高的要求。特别是，城市场景将这些努力推向了极限，因为许多快速移动的声源存在于建筑环境中，声音会反复反射和衍射。作为完成的实时衍射建模研究项目的一个成果，已经建立了一个快速的传播路径查找算法，是能够集成的反射和衍射。实时率满足使用先进的延迟线，短滤波器单元和一个新的双耳方向聚类算法，产生一个双耳可听化音频流在准恒定负载。理论上可能的传播路径和一个国家的最先进的计算机上的实际可处理的路径数之间的显着差异已被确定。假设由于a）声音来自许多入射方向和B）存在许多不同声源这两个事实，绝大多数理论传播路径变得不可感知。为了解决这个问题，必须进行关于路径重要性的评估。计划在技术和感知层面上调查传播路径的重要性。技术方法是基于声学比例模型测量和模拟结果的比较。基准案例代表了城市环境中建筑物和街道峡谷的典型情况。将模型中的变化与实际传递函数进行比较，以揭示声压级和干涉图的结果，例如，对于较高的反射和衍射级。最后，在一个听实验中，感知的影响进行了研究时，不同的重要性过滤器应用到一组理论传播路径。在这种情况下，电平阈值和可听度的着色效果的规格进行了研究ABX比较。拟议项目的主要成果是为城市情况提供一个开放的基准，作为高质量支持材料和文档（目前还没有）的参考，以及如何根据不同环境的感知重要性过滤理论传播路径的建议目录，例如在消费者台式PC上运行户外可听化。

With the research project, we want to create a broader fundament towards the auralization of urban environments. The challenge to auralize virtual acoustic environments imposes high demands on methodology and implementation of both propagation simulation and digital signal processing. In particular, urban scenarios push the efforts to the limits, as many fast-moving sound sources present in a built environment, where sound is repeatedly reflected and diffracted. As an outcome of the completed research project on real-time diffraction modeling, a fast propagation path finding algorithm has been established that is able to integrate both reflections and diffractions. Real-time rates are met by using advanced delay lines, short filter units and a new binaural direction clustering algorithm that generates a binaural auralization audio stream at quasi-constant load. A significant discrepancy between theoretically possible propagation paths and the actually processable path number on a state-of-the-art computer has been identified. The assumption is made, that the vast majority of theoretical propagation paths becomes imperceivably due to both the fact that a) sound is coming from many incident directions and b) many different sound sources are present. To solve this issue, an evaluation regarding the importance of paths must be conducted. It is planned to investigate the importance of propagation path on a technical and perceptual level. The technical approach is based on comparison of results from acoustic scale model measurements and from simulations. The benchmark cases represent typical situations with buildings and street canyons in urban environments. Variations in the modeling are to be compared with the actual transfer functions to reveal the consequence on sound pressure level and interference pattern, e.g. for higher reflection and diffraction orders. Finally, in a listening experiment, perceptual influences are investigated when different importance filters are applied to the set of theoretical propagation paths. The specification of level thresholds and audibility of coloration effects in this context are studied by ABX comparison. The main result of the proposed project is an open benchmark for urban situations as reference with high quality supporting material and documentation (which is not available at the moment) and a catalogue of recommendations how to filter theoretical propagation paths based on perceptual importance for different contexts, e.g. to run an outdoor auralization on a consumer desktop PC.