• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 Ⅳ
• /
• pp.2451-2454
• /
• 2003

In this paper, we propose the computation reduction methods of LSP(Line spectrum pairs) transformation that is mainly used in CELP vocoders. In order to decrease the computational time in real root method the characteristic of four proposed algorithms is as the following. First, scheme to reduce the LSP transformation time uses met scale. Developed the second scheme is the control of searching order by the distribution characteristic of LSP parameters. Third, scheme to reduce the LSP transformation time uses voice characteristics. Developed the fourth scheme is the control of searching interval and order by the distribution characteristic of LSP parameters. As a result of searching time, computational amount, transformed LSP parameters, SNR, MOS test, waveform of synthesized speech, speech, spectrogram analysis, searching time reduced about 37.5％, 46.21％, 46.3％, 51.29％ in average, computational amount is reduced about 44.76％, 49.44％, 47.03％, 57.40％. But the transformed LSP parameters of the proposed methods were the same as those of real root method.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2011년 춘계학술대회논문집
• /
• pp.261-265
• /
• 2011

An adaptive wavelet transformation method with high order accuracy is proposed to allow efficient and accurate flow computations. While maintaining the original numerical accuracy of a conventional solver, the scheme offers efficient numerical procedure by using only adapted dataset. The main algorithm includes 3rd order wavelet decomposition and thresholding procedure. After the wavelet transformation, 3rd order of spatial and temporal accurate high order interpolation schemes are executed only at the points of the adapted dataset. For the other points, high order of interpolation method is utilized for residual evaluation. This high order interpolation scheme with high order adaptive wavelet transformation was applied to unsteady Euler flow computations. Through these processes, both computational efficiency and numerical accuracy are validated even in case of high order accurate unsteady flow computations.

• 한국재료학회:학술대회논문집
• /
• 한국재료학회 2010년도 춘계학술발표대회
• /
• pp.39.2-39.2
• /
• 2010

Recently, hot stamping process has been paid attention greatly by automobile makers in accordance with the fuel efficiency and environmental issues as well as crash safety issue. The hot-stamped parts, however, demand extreme mechanical properties such as tensile strength of over 1470 MPa or equivalently Vickers hardness of around 450. In this work, to meet the demand efficiently, a method to predict mechanical property of hot-stamped parts based on numerical phase transformation scheme has been proposed associated with the thermo-mechanical coupled finite element analysis. This work deals with various phase transformation equations and validates them to select appropriate model for 0.2C-0.1Si-1.4Mn-0.5Cr-0.01Mo-0.002B steel sheet. The authors show that an efficient method saving time and cost to develop hot-stamped automobile parts ensuring suitable mechanical properties such as Vickers hardness and strength.

• 한국인쇄학회지
• /
• 제20권2호
• /
• pp.85-93
• /
• 2002

Recently, the flatbed scanner is widely used to input color original copy in printing industry. In the process of RGB-to-CMY transformation for the scanned colors using the look-up table(LUT) based on the color gamut of the actual output device, however, the problem that some colors in the original out of the gamut cannot be printed, may occur. In this study, another color transformation method applicable to any kind of originals using the new LUT with extended gamut based on the ideal output device, was examined and proposed.

• 한국통신학회논문지
• /
• 제34권10C호
• /
• pp.968-974
• /
• 2009

본 논문에서는 H.264의 부호화 고속 처리를 위한 skip 모드 블록의 사전 예측을 통한 변환 과정 및 양자화 과정 생략 기법에 대해 제안한다. 정수 여현 변환과 양자화의 유도 과정을 이용하여 생략 가능 블록을 예측하고, 이를 이용하여 압축 효율에 영향 없이 효과적으로 양자화 생략하는 방식에 대해 기술한다. 실험 결과를 통해 제안방식을 이용한 예측된 블록이 효과적으로 예측되어 생략되었음을 확인할 수 있었다.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2001년도 추계종합학술대회
• /
• pp.579-582
• /
• 2001

벡터변형은 벡터 양자화(VQ)와 부호화를 통합한 새로운 방법이다. 최근까지 부호화에 적용된 코드북 생성은 LBG 알고리즘이었으나 신경회로망을 기반으로 한 자기생성 특성맵(SOFM: Self Organizing Feature Map)의 장점을 이용하면 시스템의 성능을 개선할 수 있다는 점에 착안하였다. 본 논문에서는 SOFM 알고리즘을 적용한 VTC(Vector Transformation coding)코드북 생성과 LBG 알고리즘의 부호화률에 대한 결과를 비교하여 분석하였다. 벡터 양자화의 문제점은 계산의 복잡성과 코드북 생성에 있으므로 본 연구에서는 이 문제의 해결을 위해 신경망 접근법을 제안한다.

• ETRI Journal
• /
• 제35권4호
• /
• pp.714-717
• /
• 2013

Peak power reduction techniques in orthogonal frequency division multiplexing (OFDM) has been an important subject for many researchers for over 20 years. In this letter, we propose a side-information-free technique that is based on the concept of random variable (RV) transformation. The suggested method transforms RVs into other RVs, aiming to reshape the constellation that will consequently produce OFDM symbols with a reduced peak-to-average power ratio. The proposed method has no limitation on the mapping type or the mapping order and has no significant effect on the bit error rate performance compared to other methods presented in the literature. Additionally, the computational complexity does not increase.

• 한국산학기술학회:학술대회논문집
• /
• 한국산학기술학회 2010년도 춘계학술발표논문집 1부
• /
• pp.257-260
• /
• 2010

In this paper, we propose the computation reduction methods of LSP(Line spectrum pairs) transformation that is mainly used in CELP vocoders. In order to decrease the computational time in real root method the characteristic of four proposed algorithms is as the following. First, scheme to reduce the LSP transformation time uses mel scale. Developed the second scheme is the control of searching order by the distribution characteristic of LSP parameters. Third, scheme to reduce the LSP transformation time uses voice characteristics. Developed the fourth scheme is the control of searching interval and order by the distribution characteristic of LSP parameters. As a result of searching time, computational amount, transformed LSP parameters, SNR, MOS test, waveform of synthesized speech, spectrogram analysis, searching time is reduced about 37.5%, 46.21%, 46.3%, 51.29% in average, computational amount is reduced about 44.76%, 49.44%, 47.03%, 57.40%. But the transformed LSP parameters of the proposed methods were the same as those of real root method.

• Journal of Mechanical Science and Technology
• /
• 제20권2호
• /
• pp.191-196
• /
• 2006

This paper presents the application of techniques of differential transformation method (DTM) to analyze the transverse vibration of a uniform Euler-Bernoulli beam under varying axial force. The governing differential equation of the transverse vibration of a uniform Euler-Bernoulli beam under varying axial force is derived and verified. The varying axial force was extended to the more general case which was high polynomial consisted of many terms. The concepts of DTM were briefly introduced. Numerical calculations are carried out and compared with previous published results. The accuracy and the convergence in solving the problem by DTM are discussed.

• ETRI Journal
• /
• 제41권1호
• /
• pp.109-116
• /
• 2019

This paper explores new techniques that are based on a hidden-layer linear transformation for fast speaker adaptation used in deep neural networks (DNNs). Conventional methods using affine transformations are ineffective because they require a relatively large number of parameters to perform. Meanwhile, methods that employ singular-value decomposition (SVD) are utilized because they are effective at reducing adaptive parameters. However, a matrix decomposition is computationally expensive when using online services. We propose the use of an extended diagonal linear transformation method to minimize adaptation parameters without SVD to increase the performance level for tasks that require smaller degrees of adaptation. In Korean large vocabulary continuous speech recognition (LVCSR) tasks, the proposed method shows significant improvements with error-reduction rates of 8.4% and 17.1% in five and 50 conversational sentence adaptations, respectively. Compared with the adaptation methods using SVD, there is an increased recognition performance with fewer parameters.