Physical Modeling of Timpani Drums in 3D on GPGPUs
Stefan Bilbao, Craig J. Webb · Journal of the Audio Engineering Society · 2013
Powerful parallel hardware, such as general purpose graphical processing units (GPGPUs), is now becoming a viable tool in audio engineering research, with great potential in applications involving heavy computational loads. One particular area of interest is large scale audio rate physical modeling synthesis of instruments embedded in a full 3D environment. This paper explores the particular test case of the timpani drum using the finite difference time domain (FDTD) method, which is suitable for parallelization. The constituents of the timpani model, namely a membrane and cavity are introduced, as well as the coupling conditions to the acoustic field. FDTD methods are then developed, with special attention paid to implementation issues in parallel hardware. An analysis of performance on the Nvidia Tesla architecture, simulation results and sound examples are presented.