Gyeong-Ho Kim

3-D sound source reconstruction and field reprediction using the Helmholtz integral equation

Abstract A method for reconstruction of the boundary conditions of an arbitrarily shaped three-dimensional (3-D) sound source from a finite number of field pressures is presented and the associated field is repredicted by using the Helmholtz integral equation. To minimize the reconstruction error due to noise in the field pressures, a filtering method is proposed and it is verified that numerically singular values which correspond to the higher-order modes of the evanescent-like components of the source velocities are eliminated. For arbitrarily shaped sources with complex boundary conditions, the reconstruction is performed accurately when the size of each element is less than a quarter wavelength and the field measurement positions are equally or randomly distributed on the surface of a sphere in the vicinity of the source. Our technique can be used in computer-aided loudspeaker design that reconstructs loudspeaker diaphragm velocities from the finite number of field pressures and repredicts the variation of the sound field procuced by a redesigned loudspeaker cabinet on which the reconstructed diaphragm is mounted.

Sound Source Localization Methods with Considering of Microphone Placement in Robot Platform

Many different methods for sound source localization have been developed. Most of them mainly depend on time difference of arrival (TDOA) or analytic head related transfer functions (HRTF). In real implementation, since the direct path between a source and a sensor is interrupted by obstacles as like a head or body of robot, it has to be considered the number of sensors as well as their positions. Therefore, in this paper, we present the methods, which are included sensor position problem, to localize the sound source with 4 microphones to cover the 3D space. Those are modified two-step TDOA methods. Our conclusion is that the different method has to be applied in case to be different microphone position on real robot platform.

An application of speech/speaker recognition system for human-robot interaction

We will introduce a real time, robust speech/speaker recognition system for isolated word recognition using distance microphone. Applying proposed system to a robot platform, robust human-robot interaction can be established for reverberant office environments. For computational effectiveness, dynamic time warping algorithm is used for pattern matching. We select the Gamma distribution contrary to the conventional Gaussian distribution to model the probability density function of total accumulated distance. By creating reference speeches at different distances, proposed algorithm shows better speech/speaker recognition performance than the case when creating reference speeches at the same distance. Experimental results show that recognition accuracy is more than 99% by creating five reference speeches at different distances in a reverberant office environment.