• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.69-72
• /
• 2002

본 논문에서 화자공간모델 evolution에 기반한 continuous density hidden Markov model (CDHMM)의 online 적응에 대한 새로운 기법을 제안한다. 학습화자의 a priori knowledge을 나타내는 화자공간모델은 factor analysis (FA) 또는 probabilistic principal component analysis (PPCA)와 같은 은닉변수모델(latent variable model)에 의해 효과적으로 나타내어진다. 은닉 변수모델은 화자공간모델뿐아니라 CDHMM 파라메터의 ajoint prior분포를 표시함으로, maximum a posteriori(MAP)적응기법에 직접 적용되어진다. 화자공간모델의 hyperparameters와 CDHMM파라메터를 동시에 순차적으로 적응하기 위해 quasi-Bayes (QB)추정 기술에 기반한 online 적응기법을 제안한다. 연속숫자음 인식과 관련된 화자적응 실험을 통해 제안된 기법은 적은 적응데이터에서 좋은 성능을 나타내며, 데이터가 증가함에 따라 성능이 지속적으로 증가함을 보여준다.

• The Journal of the Acoustical Society of Korea
• /
• v.3 no.2
• /
• pp.46-58
• /
• 1984

본 논문은 한국어 음성 인식을 위한 유성음, 무성음, 묵음 식별에 관한 연구이다. 주어진 음성 구간을 3가지 음성 신호 부류로 식별하기 위하여 패턴 인식 방법을 사용하였다. 여기에 사용한 분석 파 라메타는 음성 신호의 영교차율, 대수 에너지, 정규화 된 첫 번째 자동 상관 계수, 선형 예측 분석에서 얻은 첫 번째 예측 계수, 그리고 예측 오차의 에너지이다. 한편 측정된 파라메타들이 다차원 가우스 확 률 밀도 함수에 따라 분산되었다는 가정하에서 어어진 최소 거리 법칙에 기본을 두고 음성 구간을 결정 하였다. 측정된 파라메타들을 여러 가지 방법으로 조합하여 식별한 결과 영교차율, 첫 번째 예측계수, 예측 오차의 에너지를 측정 파라메타로 사용했을 때 1%보다 적은 식별 오차율을 얻었다.

• The Journal of the Acoustical Society of Korea
• /
• v.3 no.2
• /
• pp.59-68
• /
• 1984

어군의 밀도를 음파로 추정할 때 어군의 총후방 산란 단면적이 어류개체의 후방 산란 단면적을 random phase로 합한 것과 같다는 가정에 대하여 실험적 검증을 시도하였다. 어체의 길이 대 음파의 파장비가 23일 때 두 마리와 세 마리의 숭어에 대하여 실시한 실험에서 전 산란각에 대한 편균이 random phase로 합할 때 보다 -1.7dB와 +1dB씩 편차를 보였다. 그러나 전체적으로는 in phase와 out phase sum 사이에서 변화함을 보였고, 최대오차는 in phase로 합해질 때로서 실제의 어군밀도 n에 비 해 최고 n\sup 2\에 이르기까지 추정 가능함을 실험결과는 보여주고 있다. 더욱이 이러한 결과는 어체간 의 간격변화에는 거의 무관하였으므로 소나무 빔폭이나 어군의 분포에 상관없이 빔폭 사이에 있는 어군 의 마리수에만 퐁후방 산란 단면적이 관계된다는 결론을 얻을 수 있었다.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.23 no.11
• /
• pp.1003-1011
• /
• 2013

A new formulation of the MNDIF method is introduced to extract highly accurate eigenvalues of concave acoustic cavities. Since the MNDIF method, which was introduced by the author, can be applicable for only convex acoustic cavities, a new approach of dividing a concave cavity into two convex domains and formulating an algebraic eigenvalue problem is proposed in the paper. A system matrix equation, which gives eigenvalues, is obtained from boundary conditions for each domain and the condition of continuity in the interface between the two domains. The validity and accuracy of the proposed method are shown through example studies.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.3
• /
• pp.162-168
• /
• 2003

The degree of diffusion, characterized by the "scattering coefficient" of surface materials, has been known to be one of the most important factors in determining the acoustical qualities of concert halls. Based on the suggested ISO method, which measures the random-incidence scattering coefficient of surfaces in a diffuse field, the scattering coefficients of different sizes and densities of wooden hemispheres and cubes were measured in model-scale reverberation rooms. As a result, wooden hemispheres with a structural depth of more than 15㎝ have the highest average (500㎐∼4㎑) scattering coefficient. It was also found that the scattering coefficient becomes higher when the diffuser density reaches about 50% for hemispheres and 30% for cubes.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.26 no.3
• /
• pp.315-322
• /
• 2016

Room acoustic properties of coupled rooms connected by an aperture has been analyzed using statistical acoustic model based on the diffused sound field assumption, which has limitation in dealing with the parameters such an room geometries and non uniform absorptivity of the boundary surfaces. In order to overcome these difficulties the acoustic diffusion model has been introduced, by which distribution of the acoustic energy density can be analyzed for various shapes and wall absorptivity. In this study acoustic properties of coupled rooms connected by an aperture(e.g. door) is analyzed using acoustic diffusion equation, which is solved numerically. The mean energy densities of two rooms obtained by the diffusion model are compared with those from the statistical model. The results show good agreement for various coupling aperture sizes and absorption coefficients. For a limiting case when the partition wall is substituted by an aperture and the two rooms eventually forms a single room, results of coupled room analysis using diffusion model show good agreement with those of a single room.

• The Journal of the Acoustical Society of Korea
• /
• v.24 no.8
• /
• pp.447-453
• /
• 2005

Laboratory measurements were performed in a uni-algae medium Cochlodinium polykrikoides (Phytoplankton, dinoflagellates) using an Underwater Ultrasound $(5\~15\;MHz)$ to study Characteristics of Acoustic Backscattering of Harmful algae. In an effort to detect the harmful algal scatterers with population density of less than 300 cells/ml that corresponds to the precaution stage of red tide, backscattered signals from various scatterer-density samples were obtained and analyzed. Correlations between volume backscattering strength (Sv) and population density (cells/ml) of scatterers in the medium have been investigated. Comparison of Volume Backscattering Strengths calculated with the fluid-sphere model [1] and the measured values showed an agreement.

• The Journal of the Acoustical Society of Korea
• /
• v.27 no.4
• /
• pp.171-177
• /
• 2008

In this paper, we have measured the fundamental properties of chime stone for Pyeongyeong. The properties are measured by wave propagation in the stone without destroying the stones, the measured properties are the chime stone density, natural frequencies of extensional wave and bending wave, and Young's modulus which is calculated by the measured properties. To find a value for Young's modulus, the fundamental frequencies which are obtained through spectrum analysis of extensional wave and bending wave are used. We calculated Young's modulus of chime stone by theoretical study and measurement on extensional wave and bending wave of the beam. As a result, we obtained Young's modulus by the fundamental frequencies of extensional wave and bending wave which deviation is within 2%.

• The Journal of the Acoustical Society of Korea
• /
• v.43 no.1
• /
• pp.112-121
• /
• 2024

The pipping system is widely used in many industries as equipment for transporting fluids over long distances. In high-pressure pipe, as the speed of the fluid increases, a loud noise is generated. Therefore, various studies have been conducted to reduce pipe noise. In this paper, a pipe noise analysis was developed to predict and quantitatively assess the flow-induced vibration and acoustic-induced vibration due to valve flow in high-temperature and high-pressure. To do this, a high-fidelity fluid analysis technique was developed for predicting internal flow in the pipe with valve. In additional, the contribution of compressible/incompressible pressure by frequency band was evaluated using the wavenumber-frequency analysis. To predict a low/middle frequency pipe noise, the vibroacoustic analysis method was developed based on Finite Element Method (FEM). And the pipe noise prediction method for the middle/high frequency was developed based on Statistical Energy Analysis (SEA).

• Proceedings of the Korea Society for Industrial Systems Conference
• /
• 1998.10a
• /
• pp.785-790
• /
• 1998

본 논문은 한국어 숫자를 연속적으로 발음한 음성의 음절 개수 검출에 관한 내용이며 음절의 최소구간 및 스펙트럼 에너지에 대한 확률밀도함수를 이용하여 연속 음성에서 음절갯수검출 알고리듬을 제안, 실험을 통하여 그 유효성을 확인하고자한다. 이를 위하여 음성자료로서는 국어 공학센터(KLE)에서 채록한 4연속 숫자음을 사용하며 음향학적 특징을 분석하기 위하여 확률밀도함수 및 음절의 최소구간 및 단위시간의 확률밀도 함수의 값을 이용하였다. 그 결과 KLE 데이터에서 스펙트럼에너지만 이용한 경우 고립음절을 3.7%이며 본 논문의 알고리듬을 적용한 경우 4음절은 약 60%의 결과가 되며 제안한 방법의 유효성을 확인하였다.