• The Journal of the Acoustical Society of Korea
• /
• v.13 no.3
• /
• pp.16-24
• /
• 1994

A single channel adaptive noise canceling (ANC) technique is presented for removing effects of additive noise on the speech signal. The conventional method obtains a reference signal using the pitch estimated on a frame basis from the input speech. The proposed method, however, gets the reference signal using the delay estimated recursively on a sample by sample basis. To estimate the delay, we derive recursion formula of autocorrelation function and average magnitude difference function. The performance of the proposed method is evaluated for the speech signals distorted by the additive white Gaussian noise. Experimental results with normalized least mean square (NLMS) adaptive algorithm demonstrate that the proposed method improves the perceived speech quality quite well besides the signal-to-noise ratio.

• Journal of Korea Society of Industrial Information Systems
• /
• v.11 no.3
• /
• pp.15-19
• /
• 2006

In shallow water active sonar environment, reverberation which is a non-white noise is one of the main source of performance degradation of target detection. In this case, the received signal is whitened before applying matched filter known as an optimum filter in the presence of white noise. However implementation of this method is very difficult because of the non-stationary characteristic of reverberation. Traditionally reverberation is assumed local stationary. In this paper, we estimate a range of stationary of reverberation signal, and then propose a pre-whitening method which improve the performance of pre-whitening block normalized matched filter in presence of non-white reverberation noise. Proposed whitener shows better whitening performance than traditional whitener because it use later as well as before reverberation of target signal. To evaluate performance of the proposed whitener, an actual measurement data sampled at the East-Sea is used for computer simulation. The target detector with new whitener is shown better performance than detector with traditional whitener.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.06e
• /
• pp.327-330
• /
• 1998

독립성분분석(ICA: Independent Component Analysis)이란 특징이 상이한 둘 이상의 신호들이 선형적으로 결합되어 있을 때 이를 효과적으로 분리하는 방법들을 통칭하며 잡음제거, 음질개선 및 신호처리 분야에서 많이 활용되고 있다. 본 논문에서는 전화음성 화자인식 시스템의 성능향상을 위해 독립성분분석을 이용하는 방법을 제안한다. 먼저 화자가 발성한 음성신호의 켑스트럼 계수를 여러 채널 함수들의 선형적인 합으로 가정하고, 독립성분분석을 이용하여 얻은 새로운 켑스트럼 벡터를 학습과 인식에 사용하였다. 실험자료는 잔화음성 화자식별기의 성능평가에 널리 쓰이고 있는 SPIDRE를 사용하였고 regodic 은닉 마코프 모델을 이용하여 문장 독립 화자식별 시스템을 구성하였다. 학습음성의 특징과 실험음성의 특징이 다른 조건에서 기존의 채널 정규화 방법들에 비해 10~15%이상 인식률이 향상되었다.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.6
• /
• pp.209-220
• /
• 2014

Compressive Sensing (CS) is a new signal acquisition paradigm which enables sampling under Nyquist rate for a special kind of signal called sparse signal. There are plenty of CS recovery methods but their performance are still challenging, especially at a low sub-rate. For CS recovery of natural images, regularizations exploiting some prior information can be used in order to enhance CS performance. In this context, this paper addresses improving quality of reconstructed natural images based on Dantzig selector and smooth filters (i.e., Gaussian filter and nonlocal means filter) to generate a new regularization called smooth residual error regularization. Moreover, total variation has been proved for its success in preserving edge objects and boundary of reconstructed images. Therefore, effectiveness of the proposed regularization is verified by experimenting it using augmented Lagrangian total variation minimization. This framework is considered as a new CS recovery seeking smoothness in residual images. Experimental results demonstrate significant improvement of the proposed framework over some other CS recoveries both in subjective and objective qualities. In the best case, our algorithm gains up to 9.14 dB compared with the CS recovery using Bayesian framework.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.28 no.2
• /
• pp.101-111
• /
• 2008

Strain imaging in a medical ultrasound imaging system can differentiate the cancer or tumor in a lesion that is stiffer than the surrounding tissue. In this paper, a strain imaging technique using quasistatic compression is implemented that estimates the displacement between pre- and postcompression ultrasound echoes and obtains strain by differentiating it in the spatial direction. Displacements are computed from the phase difference of complex baseband signals obtained using their autocorrelation, and errors associated with converting the phase difference into time or distance are compensated for by taking into the center frequency variation. Also, to reduce the effect of operator's hand motion, the displacements of all scanlines are normalized with the result that satisfactory strain image quality has been obtained. These techniques have been incorporated into implementing a medical ultrasound strain imaging system that operates in real time.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04a
• /
• pp.553-555
• /
• 2001

블라인드 등화기법은 별도의 훈련신호없이 효율적인 데이터 전송을 위한 등화기 탭 수정을 수행하는 방법이다. 본 논문에서는 이중 모드를 갖는 SCS-MMA 방법을 제안한다. 기존의 CMA와 MMA 기법들은 자승평균 오차함수(mean squared error function)를 기반으로 하는 포물선을 이루지 않는 비용함수를 사용하므로, 부적절한 국부 최소값으로 수렴할 수 있다. 제안하는 방법은 정규화된 MMA 등화 방법을 기반으로 수렴 속도의 개선과 요구되지 않은 국부 최소값으로의 수렴진행을 방지위해 SCS(soft constraint satisfaction) 알고리듬을 구현하였다. 또한, 입력 신호에 신뢰도를 주어 결정지향 알고리듬으로 자동 전환하는 방법을 적용한다. 이를 통해, 보다 빠른 수렴과 정상상태에서 결정지향 알고리듬에서와 같은 평균 오차값을 보장할 수 있다. 실험 결과 제안된 알고리듬이 기존의 방법들보다 수렴속도와 안정성에 있어 우수한 성능을 갖음을 볼 수 있다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.22 no.6
• /
• pp.681-686
• /
• 2012

This paper studied the speech parameters less affected by the human emotion for the development of the robust emotional speech recognition system. For this purpose, the effect of emotion on the speech recognition system and robust speech parameters of speech recognition system were studied using speech database containing various emotions. In this study, mel-cepstral coefficient, delta-cepstral coefficient, RASTA mel-cepstral coefficient, root-cepstral coefficient, PLP coefficient and frequency warped mel-cepstral coefficient in the vocal tract length normalization method were used as feature parameters. And CMS (Cepstral Mean Subtraction) and SBR(Signal Bias Removal) method were used as a signal bias removal technique. Experimental results showed that the HMM based speaker independent word recognizer using frequency warped RASTA mel-cepstral coefficient in the vocal tract length normalized method, its derivatives and CMS as a signal bias removal showed the best performance.

• Journal of the Institute of Convergence Signal Processing
• /
• v.12 no.1
• /
• pp.56-61
• /
• 2011

Many control techniques have been proposed in order to improve the control performance of the control system. The degree of freedom on control in the control system is decided by the number of the closed-loop transfer function which can adjust independently. The controller design scheme with two degrees of freedom(TDOF) is extensively used for securing the good control performance to trace a desired value and reject disturbance. In this paper, PID controller is used by controller with TDOF and the design method for control system with TDOF is proposed by the normal method. Using the coefficients of the transfer function of the plant, the transfer function of the control system is normalized by the proposed design method and the parameters of the controller are determined. The control system with the TDOF is constructed by using this method. Through the simulation results, the usefulness of the proposed algorithm is proved.

• Journal of Digital Convergence
• /
• v.19 no.10
• /
• pp.253-263
• /
• 2021

Machine learning is the process of constructing a cost function using learning data used for learning and an artificial neural network to predict the data, and finding parameters that minimize the cost function. Parameters are changed by using the gradient-based method of the cost function. The more complex the digital signal and the more complex the problem to be learned, the more complex and deeper the structure of the artificial neural network. Such a complex and deep neural network structure can cause over-fitting problems. In order to avoid over-fitting, a weight decay regularization method of parameters is used. We additionally use the value of the cost function in this method. In this way, the accuracy of machine learning is improved, and the superiority is confirmed through numerical experiments. These results derive accurate values for a wide range of artificial intelligence data through machine learning.

• The Journal of the Acoustical Society of Korea
• /
• v.36 no.4
• /
• pp.279-284
• /
• 2017

The localization of sources has a numerous number of applications. To estimate the position of sources, the relative time delay between two or more received signals for the direct signal must be determined. Although the GCC (Generalized Cross-Correlation) method is the most popular technique, an approach based on CCA (Canonical Correlation Analysis) was also proposed for the TDE (Time Delay Estimation). In this paper, we propose a new adaptive algorithm based on CCA in order to utilized the sparsity in the eigenvector of CCA based time delay estimator. The proposed algorithm uses the eigenvector corresponding to the maximum eigenvalue with log-sum regularization in order to utilize the sparsity in the eigenvector. We have performed simulations for several SNR(signal to noise ratio)s, showing that the new CCA based algorithm can estimate the time delays more accurately than the conventional CCA and GCC based TDE algorithms.