• The Journal of the Acoustical Society of Korea
• /
• v.28 no.8
• /
• pp.832-842
• /
• 2009

Automatic speech recognition (ASR) has been successfully applied to many real human computer interaction (HCI) applications; however, its performance tends to be significantly decreased under noisy environments. The invention of audio visual speech recognition (AVSR) using an acoustic signal and lip motion has recently attracted more attention due to its noise-robustness characteristic. In this paper, we describe our novel integration scheme for AVSR based on a late integration approach. Firstly, we introduce the robust reliability measurement for audio and visual modalities using model based information and signal based information. The model based sources measure the confusability of vocabulary while the signal is used to estimate the noise level. Secondly, the output probabilities of audio and visual speech recognizers are normalized respectively before applying the final integration step using normalized output space and estimated weights. We evaluate the performance of our proposed method via Korean isolated word recognition system. The experimental results demonstrate the effectiveness and feasibility of our proposed system compared to the conventional systems.

• The Journal of the Acoustical Society of Korea
• /
• v.16 no.4
• /
• pp.94-100
• /
• 1997

In order to investigate the radiation of sound from the King Song-Duk bell, we measured the sound pressure around the bell at every 30$^{\circ}$ using a microphone line array which was composed of 30 microphones separated by 15cm;total number of measurement point was 360. The sound field was estimated by using cylindrical acoustic holography. The spectrum of measured sound pressure demonstrates that it is almost like white noise in the very beginning, but in 10 seconds two close frequency components(64.06Hz, 64.38Hz) remain and make a famous beating. This beating sound is often believed to make unique sound of the King Song-Duk bell. The mode shapes of that frequency components are the same except that one is rotated by 45$^{\circ}$ from the other. This phenomenon occurs at the other pairs of components are the same except ones in very high frequency range where the mode shapes are rather complex.

• The Journal of the Acoustical Society of Korea
• /
• v.32 no.1
• /
• pp.22-31
• /
• 2013

In this paper, we propose a method that estimates the channel response from underwater communication signals with MMSE (Minimun Mean Squared Error) and detects dominant components automatically based on power of response components using CFAR (Constant False Alarm Rate). Statistical characteristics are analyzed with variation of magnitude and phase and time coherence via experimental data obtained by drifting transmitter and receiver. We show that bit error rate has small difference, 1.2 times, compared with the case using every channel information estimated within data period when estimation and equalization is performed with extracted characteristic obtained by the proposed method.

• Journal of Broadcast Engineering
• /
• v.23 no.4
• /
• pp.519-525
• /
• 2018

This paper deals with noise suppression of multi-channel data captured by microphone array using multi-channel Wiener filter. Multi-channel Wiener filter does not rely on information about the direction of the target speech and can be partitioned into an MVDR (Minimum Variance Distortionless Response) spatial filter and a single channel spectral filter. The acoustic transfer function between the single speech source and microphones can be estimated by subspace decomposition of multi-channel Wiener filter. The errors are incurred in the estimation of the acoustic transfer function due to the errors in the estimation of correlation matrices, which in turn results in speech distortion in the MVDR filter. To alleviate the speech distortion in the MVDR filter, diagonal loading is applied. In the experiments, database with seven microphones was used and MFCC distance was measured to demonstrate the effectiveness of the diagonal loading.

• International Journal of Fluid Machinery and Systems
• /
• v.10 no.2
• /
• pp.146-153
• /
• 2017

Sediment flow through hydropower components causes hydro-abrasive erosion resulting in loss of efficiency, interruptions in power production and downtime for repair/maintenance. Online instruments are required to measure/capture the variations in sediment parameters along with collecting samples manually to analyse in laboratory for verification. In this paper, various sediment parameters viz. size, concentration (TSS), shape and mineral composition relevant to hydro-abrasive erosion were measured and discussed with respect to a hydropower plant in Himalayan region, India. A multi-frequency acoustic instrument was installed at a desilting chamber to continuously monitor particle size distribution (PSD) and TSS entering the turbine during 27 May to 6 August 2015. The sediment parameters viz. TSS, size distribution, mineral composition and shape entering the turbine were also measured and analysed, using manual samples collected twice daily from hydropower plant, in laboratory with instruments based on laser diffraction, dynamic digital image processing, gravimetric method, conductivity, scanning electron microscope, X-ray diffraction and turbidity. The acoustic instrument was able to capture the variation in TSS; however, significant deviations were found between measured mean sediment sizes compared to values found in the laboratory. A good relation was found for turbidity ($R^2=0.86$) and laser diffraction ($R^2=0.93$) with TSS, which indicated that turbidimeter and laser diffraction instrument can be used for continuous monitoring of TSS at the plant. Total sediment load passed through penstock during study period was estimated to be 15,500 ton. This study shall be useful for researchers and hydropower managers in measuring/monitoring sediment for hydro-abrasive erosion study in hydropower plants.

• Journal of Korea Multimedia Society
• /
• v.3 no.3
• /
• pp.263-271
• /
• 2000

In this paper, an acoustic echo canceller(AEC) is implemented by ADSP-2181. This AEC uses a noise robust adaptive algorithm and a postprocessing method which attenuates residual echo using cross-correlation between estimated error signal and microphone input signal. We propose new postprocessing method that uses two thresholds to prevent signal distortion after postprocessing and to improve the performance of AEC without extra computational burden. Through experiments using a 16 bit fixed-point DSP board (ADSP-2181 EZ-KIT Lite board), it is shown that the noise robust adaptive algorithm performs well in the double-talk situations and the convergence speed is comparable to NLMS. Using the postprocessor, ERLE is improved about 20 dB. As a result, the AEC with a postprocessor shows better performance than conventional ones.

• Journal of computational fluids engineering
• /
• v.17 no.4
• /
• pp.81-86
• /
• 2012

The objective of the present study is to investigate the effects of various turbulence models on the aerodynamic noise of an air-conditioner (AC) indoor unit. The results from URANS (unsteady Reynolds-averaged Navier-Stokes) simulations with the standard k-$\varepsilon$, k-$\omega$ shear stress transport (SST) and Spalart-Allmaras (S-A) turbulence models were analyzed and compared with the noise data from the experiments. The frequency spectra of the far-field acoustic pressure were computed using the Farrasat equation derived from the Ffowcs Williams-Hawkings (FW-H) equation based on the acoustic analogy model. Two fixed fan casings and the rotating cross-flow fan were used as the source surfaces of the dipole noise in the Farrasat equation. The result with the standard k-$\epsilon$ model showed a much better agreement with the experimental data compared to the k-w SST and S-A models. The differences in the pressure spectra from the different turbulence models were discussed based on the instantaneous vorticity fields. It was found that the over-estimated power spectra with the k-w SST and S-A models are related to the emphasized small-scale vortices produced with these models.

• Fisheries and Aquatic Sciences
• /
• v.12 no.2
• /
• pp.152-161
• /
• 2009

The largehead hairtail (Trichiurus lepturus) is one of the most common fisheries stocks in the East China Sea and the Yellow Sea. The species is caught using a variety of fishing tools, such as a stow net or a long line, as well as jigging and trawling. Scientific investigations have been conducted throughout the world to enable evidence-based estimations for the management and protection of the main fisheries biomass. For example, inshore and offshore hydro acoustic surveys are performed annually using bottom- and mid-water trawls around the Korean Peninsula. However, to date, no acoustic survey has been conducted to estimate fish size distribution, which is necessary to construct a data bank of target strength (TS) relative to fish species, length (L), and frequency. This study describes the frequency and length dependence of TS among fishes in Korean waters for the purpose of constructing such a TS data bank. TS measurements of the largehead hairtail were carried out in a water tank (L 5 m$\times$width 6 m$\times$ height 5 m) at frequencies of 50, 75, 120, and 200 kHz, using a tethering method. The average TS patterns were measured as a function of tilt angle, ranging from $-45^{\circ}$ (head down) to $+45^{\circ}$ (head up) every $0.2^{\circ}$. The length conversion constant ($b_{20}$) was estimated under the assumption that TS is proportional to the square of the length. In addition, in situ TS measurements on live largehead hairtails were performed using a split beam echo sounder.

• Ocean and Polar Research
• /
• v.23 no.4
• /
• pp.425-431
• /
• 2001

From an ecological point of view, the western part of the Atlantic sector is one of the most productive areas in the Southern Ocean. Antarctic krill (Euphausia superba) and krill-dependent predators such as fish, seals, and birds are abundant there, and most krill fisheries have operated in this area since 1970s. The hottest issues for the proper management of krill resources nowadays are to determine total biomass in this area, and to identify environmental forces controlling stock fluctuation. This paper reviews and collates information on ongoing oceanographic activities in the Antarctic Peninsula region concerning these issues. To delineate the status and function of Antarctic krill population in Antarctic marine ecosystems, multinational researches along the Antarctic Peninsula area have been developing recently. Four member states of CCAMLR (Japan, Russia, UK, and USA) had conducted acoustic surveys in January-February 2000 (socalled CCAMLR-2000 survey), and krill standing stock at 120 kHz was estimated to be 44.29 million metric tonnes in the western Atlantic sector of the Southern Ocean. On the other hand, the Southern Ocean GLOBEC (SO-GLOBEC) Programme has prepared a serial winter survey to examine the factors that govern krill survivorship and distribution in relation to shelf circulation processes. Ship-based surveys using ice-breakers are being conducted by three nations (Germany, UK, and USA) around the Marguerite Bay during the austral fall and winter 2001 and 2002. In addition to these two large-scale surveys, some CCAMLR members have carried out joint oceanographic surveys near the South Shetland Islands to detect ecosystem changes since 1994. Especially from December 1999 to February 2000, in conjunction with CCAMLR-2000 survey, four nations (Japan, Korea, Peru, and USA) conducted acoustic surveys to produce time-series information on krill distribution and biomass near the South Shetland areas. Though the aims of each program and the approach to solve the scientific questions were different each other, the results from each program fill the gaps between programs. Further cooperation and exchange in these activities could be beneficial to each program.

• The Journal of the Acoustical Society of Korea
• /
• v.14 no.2
• /
• pp.40-49
• /
• 1995

Application of the filtered-x LMS adaptive filter to active noise cancellation requires to estimate the transfer charactersitics between the output and the error signal of the adaptive canceler. In this paper, we derive the filtered-x adaptive noise cancellation algorithm and analyze its convergence behavior when the acoustic noise consists of multiple sinusoids. The results of the convergence analysis of the filtered-x LMS algorithm indicate that the effects of the parameter estimation inaccuracy on the convergence behavior of the algorithm are characterized by two distinct components : Phase estimation error and estimated gain. In particular, the convergence is shown to strongly affected by the accuracy of the phase response estimate. Simulation results are presented to support the theoretical convergence analysis.