Embedded physical modeling synthesis in three dimensional environments

3D audio rendering of virtual spaces, for purposes of artificial reverberation, or in concert hall auditioning has seen great advances in recent years. Of particular interest are wave based techniques, such as finite difference time domain methods. Such methods are computationally intensive, and parallel architectures, such as GPGPUs, can be of use in accelerating computation times. A further use of such methods is in synthesis—through the embedding of physical models in a three dimensional space, allowing the complete spatial rendering of the acoustic field. In this paper, a variety of membrane- and plate-based percussion instruments will be discussed, with special emphasis on implementation issues in parallel hardware. Sound examples will be presented.