Abstract

Over the last decade the architecture of graphics accelerators (GPUs) has evolved dramatically, outpacing traditional general purpose processors (CPUs) with an average 2.25-fold increase in performance every year. With massive processing capabilities and high-level programmability, current GPUs can be leveraged for applications far beyond visual rendering.
This talk provides an overview of modern programmable GPUs and how they can be applied to acoustics and audio rendering for virtual reality or gaming applications. For tasks ranging from sound synthesis and audio signal processing to numerical acoustic simulations, GPU's massive parallelism and dedicated instructions can offer a 5–100-fold performance improvement over traditional CPU implementations. Such benefits are illustrated with results from 3D audio processing and sound scattering simulations, and future opportunities for auralization on many-core processors will also be discussed.

Bio

Nicolas Tsingos received M.Sc. and Ph.D.degrees in computer science from the Joseph Fourier University in Grenoble, France, and an advanced research degree (habilitation) from the University of Nice–Sophia Antipolis, France. He is a senior researcher at Dolby Laboratories, where he develops interactive audio-rendering technologies, primarily targeted at the gaming and cinema markets.
He is also a tenured researcher at INRIA, the French National Institute for Computer Science. Previously he was a member of the technical staff at Bell Laboratories/Lucent Technologies in Murray Hill, NJ.
His research interests include computer graphics, interactive audio simulation for virtual environments, architectural acoustics, sound synthesis, and efficient audio visual rendering. Some of the perceptual audio-rendering techniques he created have been used in commercial video games, including Test Drive Unlimited and Alone in the Dark by Eden Games/ATARI.

This lecture is part of CNMAT's spatial audio lecture series: http://cnmat.berkeley.edu/spatialaudiolectures