Improving Binaural Simulation of Structural Acoustics

Phase I showed that the Singular Value Decompostion (SVD) technique significantly reduces processing requirements for 3D binaural rendering using Head Related Transfer Function (HRTF) datasets. Equally important, the Equivalent Source Reduction (ESR) experiments completed show a dramatic simplification of the problem of acoustically modeling surfaces, with little to no perceivable difference in the auralization. While AuSIM3D? real-time binaural rendering capacity has grown from a few point sound sources four years ago to a few dozen point sources presently, largely due to gains in commodity processor speed, the performance improvement capacity shown by SVD and the simplification of the acoustic surface modeling problem demonstrated possible with ESR has been much more dramatic. Our results show hundreds of point sources can now be realistically rendered in real-time, and that these sources can represent a relatively large surface with negligible loss of accuracy. When these techniques are integrated in the AuSIM3D client and server applications as proposed in Phase II, it will be possible for the first time to binaurally render the sound of an acoustic disturbance applied to a surface model of considerable area. AuSIM is also eager to optimize the ESR and SVD implementations into commercial-quality software applications to create the next generation of AuSIM3D technology. With the newly added horsepower, AuSIM3D applications will be soon able to render directly the acoustic data generated by industry standard Finite Element Analysis (FEA) acoustics vendors such as MSC/Nastran, COMET Acoustics, and ABAQUS. These firms have already signed Letters of Intent to provide whatever technical support, collaboration, and marketing resources we need to become fully interoperable with their systems. Soon designers will be able to regularly hear the acoustic performance of the surfaces they've designed while the models are still on the electronic drawing board.