Skip to content

Smashing News

Computer scientists break into the world of synthetic sounds  Computer scientists break into the world of synthetic sounds A dinner plate falls to the ground, emitting a loud crash as it shatters into dozens of tinkling shards. This is no dining hall mishap—it's a computer simulation. For the past year, computer science professor Doug James […]

Share

Computer scientists break into the world of synthetic sounds
 

Computer scientists break into the world of synthetic sounds

A dinner plate falls to the ground, emitting a loud crash as it shatters into dozens of tinkling shards. This is no dining hall mishap—it's a computer simulation. For the past year, computer science professor Doug James and graduate student Changxi Zheng have been developing a method for digitally synthesizing the sounds of fracturing materials. They presented their work at SIGGRAPH, a major computer graphics conference, in Los Angeles in July.

Digital sound design has historically relied on repetitive, prerecorded samples, but James and Zheng hope their technology—which creates original sounds based on the physical properties of the objects in question—will someday be used in movies, video games, and training simulations to offer more distinct experiences. "We've got these compelling scenes that look visually interesting," James says of the current crop of computer-generated entertainment, "but they're basically silent movies." In addition to their fracture work, the pair has also collaborated on synthesizing the sounds of fluid interactions like splashing water, research they presented at SIGGRAPH 2009.

Their fracture algorithm first calculates the shapes of the fragments created from the breaking object; the computer then estimates each fragment's vibration based on the energy of the impact and extrapolates the sound it would create. Finally, those noises are combined to create the sound of the object being broken.

To save processing time, the algorithm takes a shortcut: it replaces the virtual fragments with simpler shapes that have similar sound properties. Still, the simulations can take several hours. James and Zheng hope that with advances in processing speed, sound designers will be using their method within five years. "Twenty or thirty years ago people started to model objects and synthesize images to look as close as possible to a real photo," Zheng says. "We have the same mission for sound."

— Marc Campasano '11

{youtube}nHH8N_lNZzI{/youtube}

Rigid-Body Fracture Sound, SIGGRAPH2010 (3:27)

Share
Share