Studies of Edge Diffraction and Scattering: Applications to Room Acoustics and Auralization

Abstract: This thesis examines ways of including edge diffraction and surface scattering to improve room acoustics auralization, i.e., the binaural replication of an acoustic environment. The approach combines numerical and psychoacoustical studies to discern what level of computational accuracy is necessary to obtain perceptually adequate replication.

In Paper I, a psychoacoustical investigation is performed on the ear's sensitivity to frequency-dependent changes in Lambert-modeled surface scattering. Using auralizations of a simulated concert hall, one finds that the frequency-dependent changes can be clearly audible over a wide frequency region and that its particular quality depends on the input signal. Frequency-dependent scattering, therefore, should be modeled, although not all auralization programs currently do this.

Paper II delves into accurate modeling of edge diffraction. Using a validated time-domain model, computations are extended to include reflection/diffraction combinations, which significantly improve agreement with scale-model measurements of a stage house. Additionally, listening tests show that coloration changes due to edge diffraction are audible even for the conservative test geometry, but that second-order diffraction to non-shadowed receivers can often be neglected. Finally, a practical implementation for binaural simulation is proposed, completing a first major step toward computing edge diffraction for more accurate auralization.

In Paper III, scattering is measured from various reflector arrays to allow study of physical diffraction phenomena and to gain new perspectives on modeling. This investigation also reveals general trends that can be of practical use in room acoustics when the wavelength is comparable to or greater than the individual panels. Results demonstrate (1) how the scattering from the array spreads over a considerably greater spatial extent than the "specular reflection zone" would predict, (2) how multi-element arrays apparently allow complex inter-panel diffraction at certain wavelengths, and (3) that the geometrical array-coverage only gives a useful estimate of the reflection strength over a limited frequency range.

Paper IV discusses two ways of improving simulation of scattering in auralization: by measurements and by computation. In addition, a hybrid model for scattering is proposed for the low- and middle-frequency ranges in room acoustics.