Hyun Jo

Observer based output feedback regulation of a class of feedforward nonlinear systems with uncertain input and state delays using adaptive gain

Abstract We consider a global regulation problem of a class of feedforward nonlinear systems with uncertain delays by output feedback. The considered system is a generalized feedforward time-delay nonlinear system in the sense that an additional uncertain constant delay exists in the main control input to the system and feedforward nonlinearity includes states and input which have uncertain time-varying delays. Moreover, when feedforward nonlinearity satisfies some restrictive condition, we extend our control problem to the case such that the delay in the main control input is also a time-varying delay. To globally regulate the considered system, we develop an output feedback controller whose gain-scaling factor involves an adaptive dynamic. Two examples are given for illustration.

Acoustic focusing headphone based on delay-and-sum beamforming

A novel headphone system with microphone arrays was proposed whose objective is to boost forward sound signal, by doing so people can communicate well with a speaker during headphone playback. The main purpose of this research is to boost the sound signal from the front, particularly for a speech signal (300 Hz~3 kHz). A delay-and-sum beamforming method is applied to an end-fire microphone array to make a proposed headphone system. Three design parameters, i.e. an aperture size, the number of microphones, and a microphone arrangement of microphone arrays are decided based on performance of beamformer with respect to each parameter. To compare the performance regarding to each design parameter, a performance measure, efficiency, was defined as ratio of two resultant sound pressures between a forward and a backward direction. A sound scattering from a head of listener is modeled by using spherical head-related transfer function (HRTF). By changing an aperture size of a microphone array, 8 cm microphone array is thought to be proper when the efficiency in speech frequency range was considered. The use of four microphones is expected as the optimum by considering both good efficiency and reducing the number of microphones. By fixing the number of microphones and an aperture size of microphone array, the efficiency was calculated in various arrangements of microphones. As a result, microphone arrangement with inter-microphone spacing of decreasing interval is selected. The design parameters such as an aperture size, the number, and an arrangement of microphones can be changed regarding to a target frequency range or a direction wanted to be boosted, and the proposed methodology can be applied similarly.

Optimization of spherical and spheroidal head model for Head Related Transfer Function customization: Magnitude comparison

Researchers have used a spherical head model to construct a HRTF (head related transfer function), which has a single adjustable parameter (radius of head) for HRTF customization mainly because of its simplicity. By using the mathematical form to calculate analytic spheroidal HRTF for an incident point source, a spheroidal head model can be used for near-field HRTF customization with two adjustable parameters (head width and head height). To see the necessity of head height consideration, two analytic solutions; spherical HRTF and spheroidal HRTF, are compared with a measured HRTF (KEMARpsilas HRTF). By varying head dimensions and ear offsets, notch frequency pattern is calculated. Through optimization procedure, we find an optimal parameters which minimizes notch pattern error between the estimated head notch pattern and notch pattern of measured HRTF. In the view point of magnitude of HRTF, performance of spherical head model and spheroidal head model is compared. Modeling error is reduced by using a spheroidal head model instead of the conventional spherical one.

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.

Confirming the perception of virtual source elevation effects created using 5.1 channel surround sound playback

Employing an array of nine speakers, five of which were at the listeners ear level, and four of which were elevated well above the listeners ear level, an experimental investigation of virtual sound source elevation was completed in each of three reproduction environments. The primary question of interest was that regarding whether the elevation of virtual sound sources could be modulated in a simple fashion using only the five ear-level speakers that form a conventional 5.1 channel surround-sound speaker layout [ITU-R. BS. 775-1. 1994]. It was found that the creation of elevated virtual sources was possible using the two surround-channel speakers or using all five ear-level speakers, and the resulting elevated virtual source imagery was compared with the source imagery associated with sound reproduced via speakers that were actually well elevated above the listeners ear level. Besides the extremely dry listening conditions of the anechoic chamber, tests were completed in a reverberation chamber, and in a moderately dry audio surround production studio, typical of the controlled acoustics featured in many critical listening spaces. The fact that similar results were observed in each of three listening environments supports the conclusion that the observed results are not idiosyncratic to any particular environment, such as the anechoic chamber, but results will likely generalize across many environments.

New 3D audio for ultra high definition digital TV; loudspeaker configuration and method for virtual elevation effect rendering

This paper suggests the next-generation audio system for ultra high definition digital TV in terms of loudspeaker layout and corresponding rendering method. First part introduces the listening test results of perceived audio quality with several loudspeaker arrangements in order to find the optimal configuration of loudspeakers for a next-generation multichannel sound system. The subjective evaluations focused on the loudspeaker configurations at the top layer were carried out with test materials by mixing in studio and from B-format recordings. The results show that the perceptual difference in the overall quality achieved with the new 10.2-channel vertical surround system with 3 top loudspeakers and the reference system was imperceptible. Second part presents the virtual elevation effect rendering algorithm which can give a listener an impression of virtual 10.2 channel speakers using the conventional 7.1 channel speaker system (ITU-R BS.775-2) placed in horizontal plane. The proposed virtual height spe...