A demonstration of the possibilities for realtime sound synthesis for large environments on today's computers.
Authors: Zhimin Ren,
Yero Yeh,
and Ming C. Lin
Email: {zren, hyeh, lin} AT cs.unc.edu
Department of Computer Science,
The University of North Carolina at Chapel Hill
We present a new contact model for physics-based sound synthesis in real-time animation and simulation. A three-level surface representation is proposed to model the rich, complex contact sounds due to object shapes, surface bumpiness, and roughness (e.g. friction). This new model can capture various types of surface interaction, including sliding, rolling, and impact, at three levels of resolutions. We demonstrate our method by generating rich, varying sounds in several interactive scenarios running in a game-like environment.
"Synthesizing Contact Sounds Between Textured Models," Zhimin Ren, Yero Yeh, and Ming C. Lin, UNC Technical Report, Chapel Hill, NC, May 2009.
Realism: A user interactively play with a three-octave xylophone. Our algorithm is able to produce the corresponding musical tones at more than 500 FPS for this complex scene, with audio generation taking 10% of the total CPU time, on a 3.4GHz Pentium-4 Laptop with 1GB RAM.
Complex Environments: Several blocks fall onto textured floors. Each box and floor are treated as aurally separate object. All the boxes and the floor are sounding. The audio simulation runs at more than 200 FPS, the application frame rate being 100 FPS. Quality Scaling ensures that the perceived sound quality does not degrade, while ensuring steady frame rates.