Modelling and validation of a radiosity-based simulation tool for sound propagation with set reflectance patterns

The acoustic perception of built environments relies on cues given by wavefronts reflecting and scattering across different surfaces. Often, surfaces in these spaces feature geometrical details or patterned textures which could noticeably impact the directivity of sound reflections and thus the soundscape. To account for this in auralizations, it is necessary to implement sound propagation simulations capable of supporting bi-directional reflectance distribution functions (BRDFs). To this end, a simulation model based on acoustic radiosity principles was developed. Acoustic radiosity models have been proven effective in the simulation of sound propagation in diffusely scattering spaces. They can be particularly meaningful in the simulation of late scattered energy in outdoor environments, since alternatives can lead to imprecise results or impractically long runtimes. Our model relies on precise numerical integration methods and bespoke architecture to support userdefined BDRFs. The current version passes several tests based on analytical models for radiosity, diffuse reflections, and classical reverberation theory. The overall validation of the model, however, relies on the comparison of measured acoustic data from real-world scenarios to the simulation output of their respective virtual analogues. Future work will look to perceptually validate of the model, by juxtaposing recorded and synthesized soundscapes from real-world urban scenarios.