This paper describes two kinds of neural networks for stereoscopic vision, which have been applied to an identification of human speech. In speech recognition based on the stereoscopic vision neural networks (SVNN), the similarities are first obtained by comparing input vocal signals with standard models. They are then given to a dynamic process in which both competitive and cooperative processes are conducted among neighboring similarities. Through the dynamic processes, only one winner neuron is finally detected. In a comparative study, with, the average phoneme recognition accuracy on the two-layered SVNN was 7.7% higher than the Hidden Markov Model (HMM) recognizer with the structure of a single mixture and three states, and the three-layered was 6.6% higher. Therefore, it was noticed that SVNN outperformed the existing HMM recognizer in phoneme recognition.

Non-singular boundary element method (BEM) codes are developed in acoustics application. The BEM code is then used to calculate unknown boundary surface normal displacements and surface pressures from known exterior near field pressures. And then the calculated surface normal displacements and surface pressures are again applied to the BEM in forward in order to calculate reconstructed field pressures. The initial exterior near field pressures are very well agreed with the later reconstructed field pressures. Only the same number of boundary surface nodes (1178) are used for the initial exterior pressures which are at first calculated by Finite Element Method (FEM) and BEM. Pseudo-inverse technique is, used for the calculation of the unknown boundary surface normal displacements. The structural object is a tuning fork with 128.4 ㎐ resonant. The boundary element is a quadratic hexahedral element (eight nodes per element).

Ryu et al. recently proposed a multiple target angle-tracking algorithm without a data association problem. This algorithm, however, shows the degraded performance on evasive maneuvering targets, because the estimated signal subspace is d,:graded in the algorithm. In this Paper, we proposed a new algorithm, in which VFF-PASTd (Variable Forgetting Factor PASTd) algorithm is applied to Ryu's algorithm to effectively handle the evasive target tracking with better time-varying signal subspace.

We propose new time-frequency (TF) tools for analyzing linear time-varying (LTV) systems and nonstationary random processes showing hyperbolic TF structure. Obtained through hyperbolic warping the narrowband Weyl symbol (WS) and spreading function (SF) in frequency, the new TF tools are useful for analyzing LTV systems and random processes characterized by hyperbolic time shifts. This new TF symbol, called the hyperbolic WS, satisfies the hyperbolic time-shift covariance and scale covariance properties, and is useful in wideband signal analysis. Using the new, hyperbolic time-shift covariant WS and 2-D TF kernels, we provide a formulation for the hyperbolic time-shift covariant TF symbols, which are 2-D smoothed versions of the hyperbolic WS. We also propose a new interpretation of linear signal transformations as weighted superposition of hyperbolic time shifted and scale changed versions of the signal. Application examples in signal analysis and detection demonstrate the advantages of our new results.

Kuo et al proposed an on-line method for an adaptive prediction error filter for improving secondary path modeling performance in the modeling method of the secondary path. This method have some disadvantages, namely having to use additive noise with the result that noise control performance is not good since it is focused on the estimated performance of the secondary path. In this paper, we proposes a modified Kuo model using gain control parameter and delay. It uses a reference signal for additive noise to improve the problems in the existing Kuo model.

The effects of linear acoustic channels have been analyzed and compensated at mel-frequency feature domain. Unlike popular RASTA filtering our approach incorporates separate filters for each mel-frequency band, which results in better recognition performance for heavy-reverberated speeches.