The Wave-based Room Acoustics Modeling (WRAM) project has been concerned with bringing the latest scientific advances in the area of virtual room acoustics modelling to the marketplace. The idea is to render the acoustic behaviour of a specified space – such as a cathedral, concert hall, or even more routine areas such as workplace environments – to a degree which is perceptually transparent to the listener. The applications are many: to building and architectural acoustics, virtual reality, and computer game audio.
Current approaches to virtual room acoustics modelling rely on simplifications of the theory of acoustic wave propagation to geometrical acoustics (e.g. ray tracing). Though common in computer graphics applications, in acoustics, such simplifications neglect many important features, such as the diffraction of sound waves around objects or corners. A complete model of virtual acoustics must retain the wave nature of the underlying theory. In practice, this required a detailed volumetric model of the space itself, down to a resolution of approximately one centimetre. Computational costs are quite high, but within the realm of possibility on modern parallel computers. In WRAM, the world’s first general audio-rate room acoustics simulation system has been put into place, operating on a GPU server.
The system is now operational as a web-based service, accepting input data in the form of standard 3D room models (Sketchup, e.g.) and produces room impulse responses, suitable for analysis of proposed architectural designs, or auralisations (virtual walkthroughs). A spinout company, Roomerical, has been registered, and will launch in October, 2018. Clients will include major multinational building acoustics and architecture firms.
