Programme
Thursday 12th February
Session Chair:
Tapio Lokki, Department of Media Technology, Helsinki University of Technology, Finland
Paper Session 5: Advanced Spatial Audio & Reverb
5-1: Precomputing Geometry-Based Reverberation Effects for Games
Nicolas Tsingos, Dolby Laboratories, San Francisco, CA, USA
Current games either pre-render reverberation effects into the sound effects or implement them at run-time using artificial reverberation filters. While interactive geometrical approaches can be used for more accurate acoustical modeling, the increased authoring complexity and the additional cost of geometrical calculations still appears to overshadow their potential benefits. This paper presents solutions to integrate off-line geometrical acoustic modeling in game environments. By precomputing image-source gradients for early reflections and directional decay profiles, we can generate location-dependent reverberation effects without storing or accessing the actual geometry at run-time. We render such reverberation effects using a frequency-domain scalable processing approach. In this context, we introduce an efficient prioritization scheme and evaluate alternative transforms for late reverberation processing. Our pipeline enables fine-grain rendering of distance and surface proximity effects and modeling of both outdoor and coupled indoor spaces with arbitrary reverberation decay profiles.
5-2: Generating a Spatial Average Reverberation Tail Across Multiple Impulse Responses
Rebecca Stewart, Mark Sandler, Queen Mary, University of London, London, UK
Auralization using impulse responses rendered or measured in non-real-time is often limited to static sound source and receiver positions. An increasingly common approach is to use a large database of previously rendered impulse responses to simulate moving through a modeled space. Described here is a method to remove redundant information from that database of impulse responses. In particular, the late reverberation tail is analyzed and a generalized version is synthesized.
5-3: Acoustic Impulse Response Interpolation for Multichannel Systems Using Dynamic Time Warping
Claire Masterson, Gavin Kearney, Frank Boland, Trinity College Dublin, Dublin, Ireland
We present a method of interpolation of room acoustic impulse responses for reducing measurement sets in multichannel convolution systems. The method employs the use of Weighted Dynamic Time Warping for the synthesis of early reflections and critical band analysis for diffuse decay decorrelation. An objective study of the reduced data sets in comparison to full spatial resolution measurements taken in a reverberant environment is presented through binaural measurements. Listening tests are conducted under controlled conditions for investigation into the perceptual attributes associated with the synthesized audio.
5-4: An Improved Parametric Model for Perception-Based Design of Virtual Acoustics
Christian Borss, Rainer Martin, Ruhr-Universitat Bochum, Bochum, Germany
In this paper we present an architecture for the generation of artificial acoustics that can be used for plausible as well as artistic virtual environments. In our approach, we create synthetic acoustics derived from few acoustical design parameters like the room geometry, the frequency dependent reverberation time, and the echo density profile. We describe how the intended design parameters are mapped to our signal generation architecture and how audio signals can be rendered for presentation with an arbitrary number of loudspeakers or with headphones.