• 한국정보전자통신기술학회논문지
• /
• 제14권4호
• /
• pp.247-256
• /
• 2021

본 논문에서는 피치 하모닉 움직임 예측과 적응적 신호 크기 예측을 이용한 패킷 손실 알고리즘을 제안한다. 스펙트럼 움직임 예측 방법은 사용 가능한 이전 패킷의 스펙트럼 상의 움직임을 일정한 부대역으로 나누어 손실된 신호의 움직임을 예측하여 복원한다. 제안하는 알고리즘에서는 음성신호를 유성음과 무성음으로 구분하여 유성음의 경우 피치 주파수를 활용하여 피치 하모닉으로 나누어 손실된 신호의 피치 하모닉 움직임을 예측하여 복원하고 무성음의 경우 스펙트럼 움직임 예측 방법을 사용하여 신호를 복원한다. 음성 프레임의 연속 손실이 발생한 경우 LMS(Least Mean Square) 예측기를 사용하여 이전 프레임의 이득 정보를 활용하여 신호 크기를 예측하여 출력 신호의 이득을 조절하는 방법을 제안한다. 객관적 평가방법인 PESQ (Perceptual Evaluation of Speech Quality) 시험을 통해 제안된 알고리즘의 성능을 평가하였고 기존의 방법보다 MOS 0.1의 성능 개선을 보였다.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 1989년도 봄 학술발표회 논문집
• /
• pp.5-10
• /
• 1989

In this study, the basic concept of design response spectrum is briefly revi-ewed. To generate the artificial earthquake acceleration, the method of superpo-sition of cosine waves is used. Theoretical developments using F.F.T. and spect-ral density function are compared. The amplitude was derived by use of the peak factor and the phase angle is d-erived by use of Monte Carlo simulation. To smoothen the match, the calculated pseudo velocity respon-se spectrum is compared with input pseudo velocity response spectrum at a set of control frequencies. With the modified spectral density function, a new acceleration and pseudo velocity response spectrum are generat-ed.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 추계종합학술대회 논문집(4)
• /
• pp.203-206
• /
• 2000

Since the amplitude of voiced fall off at about -20dB/decade, dynamic range is often compressed prior to spectral analysis so that details at weak, high frequencies may be visible. Preemphasizing the speech, either by differentiating the analog speech s$\sub$a(t) prior to A/D conversion or by differencing the discrete-time s(n)=s$\sub$a(nT), compensating for falloff at high frequencies. The most common form of preemphasis is y(n)=s(n)-As(n-1), where A typically lies between 0.9 and 1.0 and reflects the degree of pre-emphasis. In This paper, we proposed that A is adjusted at each time by measuring the slope of envelope in frequency domain.

• 한국음향학회:학술대회논문집
• /
• 한국음향학회 2000년도 하계학술발표대회 논문집 제19권 1호
• /
• pp.31-34
• /
• 2000

본 연구에서는 부가잡음에 의한 음성신호의 왜곡에 대해 다양한 음성개선 기법을 전처리기로 도입하여 HMM(Hidden Markov Model)에 기반 한 음성인식 시스템의 인식성능을 평가하였다. 음성개선 기법으로는 MMSE(Minimun Mean Square Error) STSA(Short-Time Spectral Amplitude Estimator) 기법과 웨이브렛 영역에서의 UWD(Undecimated Wavelet Denoising), CWD(Conventional Wavelet Denoising) 기법을 적용하였다. 잡음이 없는 데이터로 훈련한 음성인식시스템에 잡음음성을 입력할 때 각 음성개선기법을 전처리기로 사용하여 신호대잡음비(Signal to Noise Ratio)에 따른 인식 성능을 비교하였다.

• 대한기계학회논문집
• /
• 제6권4호
• /
• pp.346-352
• /
• 1982

The vibration characteristics of hyoid gears are experimentally investigated with backlash and tooth mesh frequency varied. Experiments are performed with two types of gear sets; well-machined and over-lapped gears. Through the experimental works, a constant torque 0.8kg.m is applied to th gears and a pulse type of trigger signal synchronizing tooth contact is used to measure the tooth mesh frequency and to average the gear vibration signal in the time domain. The time averaged vibration signals are frequency analyzed by a digital signal analyzer. The experimental results show the vibration characteristics of the two different types of gears in relation to backlash amount and mesh frequency variations are well differentiated by harmonic amplitudes and their ratios, and peak-to-peak amplitude of the gear vibration.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2002년도 11월 학술대회지
• /
• pp.187-190
• /
• 2002

In this paper, a speech enhancement system using microphone array with MMSE-STSA (Minimum Mean Square Error-Short Time Spectral Amplitude) estimator based post-processing is proposed. Speech enhancement is first carried out by conventional delay-and-sum beamforming (DSB). A new MMSE-STSA estimator is then obtained by refining MMSE-STSA estimators from each microphone, which is applied to the output of conventional DSB to obtain additional speech enhancement. Computer simulation for white and pink noises show that the proposed system is superior to other approaches.

• Annals of Clinical Neurophysiology
• /
• 제1권1호
• /
• pp.47-54
• /
• 1999

The Doppler in neurosonologic examination could be applied to blood flow to determine its movement, the direction of its movement, and how fast it is. Indicies of the Doppler study denoted velocity, direction, and amount of RBC in the examined vessel. Systolic. diastolic, and mean blood flow velocities represent velocity of RBCs in a sample volume. Blood flow direction to the probe means direction of RBC to the probe. Size of amplitude displays toe amount of the RBCs passing the sample volume. Spectral broadening means presence of turbelence. The RBC movements and hemodynamics at the examined vessels can be estimated by analysis of Doppler indicies The formation and meaning of each of neurosonologic Doppler study is described in the present review.

• 대한기계학회논문집
• /
• 제7권1호
• /
• pp.36-45
• /
• 1983

In order to characterize the machine tool feed-drive dynamics, thread cutting experiments are performed with cutting conditions and slide-way lubrication varied. During the experiments, the carriage, tool post and tail stock accelerations in the feed direction are measured, and analyzed by employing the spectral analysis method. It is found that the tool post vibration in the feed direction during thread cutting operation is mainly due to those of the carriage and the workpiece. Other structure-related vibrations show little effects on the tool post vibration. The characteristics of the carriage vibration is shown to be fairly consistent, except the vibration amplitude, regardless the variations in cutting condition and lubrication within the experimental range. The experimental results suggest that the feeddrive system can be modelled as a 2 DOF damped oscillatory system.

• Journal of Biosystems Engineering
• /
• 제22권3호
• /
• pp.269-278
• /
• 1997

An automatic level measurement system was developed to level the land fer direct seeding of rice. A laser transmitter/receiver set was used to measure land-level. The inclination error occurred in level measurement on irregular land surface could be compensated by attaching rotating mass. The level measuring experiments were performed on three kinds of different shapes(step, random, sine). This system could accurately measure step level of which amplitude was 40mm in 0.5s, random level change within $\pm$ 5mm maximum measurement error, and sine level change of which spatial frequency was 0.5m-1. To verify performance of the inclination error compensation system, frequency transfer function(acceleration input vs. inclination error) was computed by spectral analysis. The inclination error was decreased about 20㏈ by error compensation system.

• Transactions on Electrical and Electronic Materials
• /
• 제15권4호
• /
• pp.182-188
• /
• 2014

When the windowed DFT algorithm is applied in harmonic measurements, the problem of initial-phase sensitivity will be encountered, this has an effect on harmonic amplitude accuracy. In this paper, the origin of initial-phase sensitivity is analyzed and the main factors that influence the level of initial-phase sensitivity are demonstrated. A method of reducing initial-phase sensitivity is proposed to increase the stability of harmonic measurements. We found that initial-phase sensitivity is determined by the side lobe peak level of the window functions when synchronous deviation is fixed. In addition, increasing the length of the time recorded can be used to remove initial-phase sensitivity. The correctness and validity of our conclusions have been confirmed through numerical results and field tests.