• 한국컴퓨터정보학회논문지
• /
• 제10권6호
• /
• pp.47-54
• /
• 2005

대부분의 화자인식 시스템은 음성 분석을 통해 화자의 특징을 음향 파라미터 형태로 추출하여 화자의 표준패턴을 만든 후, 입력된 미지의 음성패턴과의 차이를 계산하여 허용 여부를 최종적으로 판단한다. 화자인식에 사용하는 파라미터는 화자의 특징을 충분히 표현함과 더불어 발성 시마다 변동이 작은 것이 바람직하다. 따라서 본 논문에서도 이를 위해서 다음과 같이 제안하였다. 벡터 양자화모델에 비잡음 환경에 강한 스펙트럼 특징과 잡음 환경에 강한 운율정보를 화자인식 시스템에 이용할 것을 제안하였다. 훈련과정에서 코드북 형성시 실제 데이터를 스펙트럼 특징과 운을 특징을 조합하여 원하는 모델 수만큼 만들었다. 인식과정에서는 입력된 테스트패턴을 각 모델간에 거리 측도로 비교하여 가부를 결정하였다. 실험결과 스펙트럼 특징과 운을 특징을 각각 이용할 경우 보다 좋은 인식율을 얻었으며, 특히 잡음 환경에서 안정된 인식율을 확보하므로 상용화의 가능성을 한층 높였다.

• 한국음향학회지
• /
• 제31권8호
• /
• pp.523-531
• /
• 2012

최근 친 환경 발전에 관한 관심이 높아지면서 태양광, 풍력, 조력, 조류 발전에 대한 수요가 점차 증가되고 있다. 이러한 친 환경 발전 방식 중 조석운동에 의한 해수의 흐름을 이용하는 조류발전은 지형적 특성에 의해 강한 조류가 발생하는 지역에서만 사용할 수 있는 특수한 발전 방식이다. 울돌목 해역은 조류 발전이 가능한 매우 강한 조류가 형성되는 지역으로 다른 지역의 해역과 구분되는 특별한 환경을 제공한다. 하지만, 해양에서의 인간의 활동은 수중소음을 야기하여 해양환경에 큰 영향을 미친다. 최근에는 환경영향평가의 중요성이 대두되면서 발전소 가동 시의 소음 특성 분석 및 전파양상 예측에 대한 필요성이 증가하고 있다. 본 연구는 조류발전소의 수중소음 특성을 측정하고 울돌목 해역으로의 전파양상을 모델링 하였다.

• 대한기계학회논문집
• /
• 제18권5호
• /
• pp.1253-1263
• /
• 1994

The experimental study is carried out to identify the combustion generated noise mechanism in free turbulent jet diffusion flames. Axial mean fluctuating velocities in cold and reacting flow fields were measured using hot-wire anemometer and LDv.The overall sound pressure level and their spectral distribution in far field with and without combustion were also measured in an anechoic chamber. The axial mean velocity is 10-25% faster and turbulent intensities are about 10 to 15% smaller near active reacting zone than those in nonreacting flow fields. And sound pressure level is about 10-20% higher in reacting flow fields. It is also shown that the spectra of the combustion noise has lower frequency characteristics over a broadband spectrum. These results indicate that the combustion noise characteristics in jet diffusion flames are dominated by energy containing large scale eddies and the combusting flow field itself. Scaling laws correlating the gas velocity and heat of combustion show that the acoustic power of the combustion noise is linearly proportional to the 3.8th power of the mean axial velocity rather than 8th power in nonreacting flow fields, and the SPL increases linearly with logarithmic 1/2th power of the heat of combustion.

• 한국소음진동공학회논문집
• /
• 제14권12호
• /
• pp.1268-1278
• /
• 2004

Measurement of noise is not only to know the information of acoustic pressure but to assess human response to noise. To find human response to transportation noise through the laboratory study we have to measure and reproduce noise. The method of noise reproduction is largely divided into monaural and binaural techniques. But human fundamentally hears sound through both ears, referred as binaural hearing. Binaural signal is different from monaural signal because it includes more information of physical phenomena like acoustical reflection, diffraction and refraction. Especially head and pinna play an important role in perceiving change of signal origin. So, the amplitude of binaural signal is higher than that of monaural signal and spectrum of both signals is discriminated. Most of assessment and regulation of transportation noise are, however, based on monaural measurement techniques. The quantitative difference between monaural and binaural measurement is investigated in this study. Comparison on several transportation noisesshows defect of information in monaural measurements.

• 한국군사과학기술학회지
• /
• 제13권4호
• /
• pp.528-533
• /
• 2010

This paper describes a calculation method of source level of a ship transient noise, which is one of the important elements for the ship detection. Aim of transient noise measurements is to evaluate of acoustic energy due to singular occurrence, which is therefore defined as non-periodic and short termed events like an attack periscope, a rudder and a torpedo door. In generally, in the case of randomly spaced impulse, the spectrum becomes a broadband random noise with no distinctive pattern. Therefore, frequency analysis is not particularly revealing for type of signal. In the paper, it is performed in time domain to analyze a transient noise. However, a source level of transient noise is required an investigation for multiple frequency band. So, in order to calculate a source level of transient noise, a design of exponential weighting function, convolution, band pass filtering, peak detection, root mean square, and parameter compensation are applied. The effectiveness of this calculation scheme is studied through computer simulations and a sea test. Furthermore, an application of the method is applied in a real case.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2006년도 추계학술대회논문집
• /
• pp.498-501
• /
• 2006

When an object or objects, rigid or flexible, presents in incident sound field, the sound wave is scattered. This, we call, is scattered sound field. It, of course, depends on the amplitude and the direction of the incident sound field as well as the geometry and the surface impedance of the scatterer(object). This paper addresses the way to measure scattered sound field by using arbitrary incident sound wave. This means that the method can decompose the scattered field from measured sound field with respect to any magnitudes and directions of incident plane-waves.

• 대한기계학회논문집B
• /
• 제24권10호
• /
• pp.1326-1334
• /
• 2000

The oscillatory excitation with a Strouhal number of 2.65 ncar the stagnation zone of hemispherical nose model was employed to control the laminar separation bubble and the transition to turbulence. The effects of oscillatory excitation upon the separation bubble and the transition were addressed in terms of kurtosis/skewness and time-frequency analyses. The measured noise spectrum of radiated sound from the turbulent boundary layer on the axi-symmetric infinite cylinder is compared with that by Sevik's wave-number white approximations. The noise sources in TBL on axi-symmetric cylinder and the caling of their far-field sound are also discussed.

• 한국소음진동공학회논문집
• /
• 제24권6호
• /
• pp.483-492
• /
• 2014

Uncomfortable feelings of occupants by indoor floor impact noise in a residential building are not accurately represented by the floor impact noise from a standard impact source. It is due to the characteristics of standard impact sources, which are different from the impact forces produced by occupants. It varies significantly by impact source, and it is not easy to be replicated for testing. As a result, the indoor floor impact noise under different acoustic conditions cannot be directly compared. Using frequency response function(FRF), which represents the input-output relationships of a dynamic system, it is possible to examine the characteristics of the system. Especially, FRF can predict the response of a linear dynamic system subjected to various excitation. To determine the relationship between impact force and the corresponding response of dynamic system in residential building, the acceleration response of a concrete slab and the floor impact noise in the living room, produced by bang-machine and rubber-ball excitation, were measured. The test results are compared to the estimates based on FRF and impact force spectrum.

• 해양환경안전학회지
• /
• 제25권6호
• /
• pp.765-771
• /
• 2019

선박 및 수중구조물의 고속, 대형화 및 요구조건 강화의 추세에 따라 유동소음 예측기술의 중요성이 강조되고 있다. 항공, 철도 등의 공력소음 분야에서는 음향상사법을 이용하여 순음 및 광대역 유동소음에 대해 활발히 연구되고 있는 반면 조선해양분야에서는 수중추진기의 날개주파수소음에 대해서만 일부 고려되고 있다. 본 논문에서는 날개면 형상의 주요 유동소음발생 메커니즘 뒷날소음을 고려 가능한 FW-H Formulation 1B를 이용하여 수중추진기 및 선저부가물의 기초요소인 수중익에 대해 광대역소음 예측기법을 연구하였다. 기존의 FW-H Formulation 1B는 공기 중의 압력상관관계 모델에 기반하여 구성되어 있어 매질에 대한 일반성 및 정확도의 한계를 가지므로 수중환경에 대해 일반성을 가지는 벽면변동압력 모델로 확장하는 방법론을 제시하였다. 공기 중 날개면의 소음계측결과와 비교해 벽면변동압력 모델을 이용할 경우 기존모델의 해석결과 대비 5 dB 이내의 오차로 정확도 관점에서의 유용성을 확인할 수 있었으며 전산유체역학과 벽면변동압력을 이용한 수중환경의 광대역소음해석 절차를 확립하고 수중익의 광대역소음 예측을 수행하였다.

• 한국소음진동공학회논문집
• /
• 제22권7호
• /
• pp.626-633
• /
• 2012

In this paper, sound pressure sensitivity of the fiber optic acoustic sensor according to sensor direction and mandrel material were investigated experimentally. Three different directions were selected as stand, lay, and hole. Hollow cylinder type mandrel dimension is 30 mm in outer diameter, 45 mm in length, and 2 mm in thickness, and about 50 m optical fibers were wounded on the surface of the mandrel. Non-directional sound speaker was used as a sound source. Sagnac interferometer and single mode fiber, a laser with 1,550 nm in wavelength, $2{\times}2$ coupler were used. Based on the experimental results, lay direction's sensitivity is the highest in the frequency range of 2 kHz~4 kHz. 'PTFE+carbon' material is more sensitive than PTFE in the frequency range of 5 kHz~20 kHz. Sound pressure detection sensitivity depends on the mandrel direction and material under certain frequency.