-----------------------------------------------------------------------
                   B O S T O N   U N I V E R S I T Y
                      Computer Science Department
                          C O L L O Q U I U M
                      Wednesday, April 17, 3:00 PM 
                       (Coffee served at 2:45 PM)
                         Seminar Room / MCS 135


                       Creating Virtual Audio Spaces 
               via Scientific Computing and Computer Vision


                             Ramani Duraiswami
                  Institute for Advanced Computer Studies 
                   University of Maryland, College Park

                                Abstract


Our recent research has focused on the creation of virtual audio scenes that
are perceptually convincing. Humans are remarkably adept at processing the
sound they receive at their two ears to perceive the spatial location (and
other attributes) of sources in an acoustic scene. While cues that arise
from the differences in level and time of arrival of the sound at the two
ears (binaural cues) are important, they are incomplete. Cues that arise
from the scattering of sound off a persons body (external ears, head, and
torso), and off the environment are also essential in explaining this
ability. These cues are determined by the location of the source, the shape
of the scattering surfaces, and the frequency content of the source. 

To recreate an audio scene virtually these scattering based cues must be
reintroduced. The cues that arise due to scattering off the listener are
encoded in the so-called "Head Related Transfer Function" (HRTF). The HRTF
exhibits considerable variation from person to person. Hitherto, the HRTF
had to be measured individually, making it difficult to use it to render
perceptually convincing spatially distributed audio. We have proposed a
novel solution to this problem, in which the individual's HRTF is computed
using numerical methods, that use as input accurate 3D surface models of a
person's ears, head and  torso,(obtained using computer vision), and fast
algorithms for the solution of the wave equation.  

The support of the National Science Foundation is gratefully acknowledged.


Host: Margrit Betke
-------------------------------------------------------------------------
For colloquium info, including directions, see http://cs-www.bu.edu/colloquium
-------------------------------------------------------------------------