DAGA 2010 - Deutsche Gesellschaft für Akustik eV

Programm DAGA 2010 141
the use of the simple perceptual coder limited the significance of the data.
Further research on combining the new model with a state-of-the-art
coder is indicated.
Mi. 18:10 Gauß B 554 Räumliche Wahrnehmung
Real-time Rendering of Moving Sound Source in a Room Based on
Principal Components Analysis
D. Rao und Y. Liu
Acoustic Lab., Physics Dept., South China Univ. of Technology
In free-field situation, convoluting sound signal with a pair of headrelated
transfer function can render a sound source virtually. In a room,
however, because there are many image sources generated by reflections
from wall, the situation is more complex, especially when the source
is moving. Rendering such scene requires 2*N (image source numbers)
times of convolution computations so as to bring difficulty in the realtime
implementation. In this paper, a method using principal components
analysis (PCA) is proposed to reduce the rendering computation load. In
this method, HRTFs were expressed as weighted sum of a set of basics
functions which can be derived from PCA. Therefore, the HRTFs of real
source and image sources at different directions share a common set of
basics function. In such way, computation reduced to 2*L(basics function
number) times of convolution computations. We used L=10 basics
function in our implementation, which was much less than the number
of image sources (36 for considering second-order image sources in
rectangular room). Objective analysis and subjective listening test validated
the proposed method. The performance and details of real-time
implementation using VC++ of the proposed method on PC were also
discussed in this paper.
Mi. 8:30 Beuth-Saal Raumakustik - Simulation und Evaluation II
The Relation Between Perception of Room Acoustics and Objective
Parameters as Calculated Using a Binaural, Nonlinear Model
J. van Dorp Schuitman und D. de Vries
Delft University of Technology
At TU Delft a new method was proposed for evaluating parameters related
to room acoustical quality. The method is based on a binaural, nonlinear
auditory model which is capable of predicting perceptual reverberance,
clarity, apparent source width and listener envelopment from
binaural audio recordings. This way the parameters are calculated in a
content-specific way and furthermore there is no need to perform impulse
response measurements. From listening test results it follows that
these parameters correlate satisfactorily with room acoustical perception,
whereas conventional acoustical parameters often do not. In this
paper the method is discussed and the relation between the model outputs
and the results of the listening tests is evaluated.