• The Journal of the Acoustical Society of Korea
• /
• v.42 no.4
• /
• pp.345-356
• /
• 2023

Target signal of passive sonar shows narrow band harmonic characteristic with a variation in intensity within a few seconds and long term frequency variation due to the Lloyd's mirror effect. We propose a signal classification algorithm based on Gated Recurrent Unit (GRU) that learns local and global time series features. The algorithm proposed implements a multi layer network using GRU and extracts local and global time series features via dilated connections. We learns attention mechanism to weight time series features and classify passive sonar signals. In experiments using public underwater acoustic data, the proposed network showed superior classification accuracy of 96.50 %. This result is 4.17 % higher classification accuracy compared to existing skip connected GRU network.

• Phonetics and Speech Sciences
• /
• v.4 no.3
• /
• pp.43-49
• /
• 2012

This paper describes the categorical effects of pitch accent contrasts in a mimicry task. It focuses, specifically, on examining how fundamental frequency (f0) variation reflects phonological contrasts from speakers of two distinct varieties of Korean (i.e., North Kyungsang and South Cholla). The results showed that, in a mimicry task using synthetic speech continua, there was a categorical effect in f0 peak timing for North Kyungsang speakers, but the timing of f0 peaks and valleys in the responses of South Cholla speakers was more variable, presenting a gradient or non-categorical effect. Evidence of categorical effects was represented as the shift of f0 peak times along an acoustic continuum for North Kyungsang speakers. The range for the shift of f0 valley times was much narrower, compared to that of f0 peak times. The degree of a shift near the middle of the continuum showed variability across individual mimicry responses. However, the categorical structure in mimicry responses regarding the clustering of f0 peak points was more significant for North Kyungsang speakers than for South Cholla speakers. Additionally, the finding of the current study implies that the location of f0 peak times depends on individuals' imitative (or cognitive) abilities.

• Korean Journal of Optics and Photonics
• /
• v.17 no.4
• /
• pp.366-370
• /
• 2006

An optical microphone incorporating a reflective diaphragm and a fiber-optic head was demonstrated. The diaphragm was made of a micromirror membrane which is suspended by a silicon bar connected to a frame, allowing fer a displacement induced by acoustic waves. A compact, simple optical head was implemented by exploiting a single multimode fiber. For the assembled microphone, the static characteristics were investigated to find the operation point defined as the optimum distance between the head and the diaphragm, and a flat frequency response with a variation of $\sim$2dB for the range of up to 2 kHz was accomplished.

• Phonetics and Speech Sciences
• /
• v.3 no.4
• /
• pp.71-81
• /
• 2011

This paper investigated how prosodic position and word type affect the phonetic structure of Korean coronal stops. Initial segments of prosodic domains were known to be more strongly articulated and longer relative to prosodic domain-medial segments. However, there are few studies examining whether the properties of prosodic domain-initial segments are affected by the information content of words (real vs. nonsense words). In addition, since the scope of domain-initial effect was known to be local to the initial consonant and the effects on the following vowel have been found to be limited, it is thus worth examining whether the prosodic domain-initial effect extends into the vowel after the initial consonant in a systematic way across different prosodic domains. The acoustic properties of Korean coronal stops (lenis /t/, aspirated /$t^h$/, and tense /t'/) were compared across Intonational Phrase, Phonological Phrase and Word-initial positions both in real and nonsense words. The durational intervals such as VOT and CV duration were cumulatively lengthened for /t/ and /$t^h$/ in the higher prosodic domain-initial positions. However, tense stop /t'/ did not show any variation as a function of prosodic position and word type. The domain-initial lenis stop showed significantly longer duration in nonsense words than in real words. But the prosodic domain-initial effect was not found in the properties of F0 and [H1-H2] of the vowel after initial stops. The present study provided evidence that speakers tend to enhance speech clarity when there is less contextual information as in prosodic domain-initial position and in nonsense words.

• Journal of Korea Multimedia Society
• /
• v.21 no.10
• /
• pp.1195-1202
• /
• 2018

Artificial neural network is widely used for its excellent performance and implementability. However, traditional neural network needs to learn the system from scratch, with the addition of new input data, the variation of the observation environment, or the change in the form of input/output data. To resolve such a problem, the technique of transfer learning has been proposed. Transfer learning constructs a newly developed target system partially updating existing system and hence provides much more efficient learning process. Until now, transfer learning is mainly studied in the field of image processing and is not yet widely employed in acoustic data processing. In this paper, focusing on the scalability of transfer learning, we apply the concept of transfer learning to the problem of guitar chord classification and evaluate its performance. For this purpose, we build a target system of convolutional neutral network (CNN) based 48 guitar chords classification system by applying the concept of transfer learning to a source system of CNN based 24 guitar chords classification system. We show that the system with transfer learning has performance similar to that of conventional system, but it requires only half the learning time.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.2
• /
• pp.221-225
• /
• 2008

In this paper, to avoid the frequency analysis requiring a high sampling rate, time-warped similarity measure algorithms, which are able to classify objects even with a low-rate sampling rate as time- series methods, are presented and proposed the DTW-Cosine algorithm, as the best classifier among them in wireless sensor networks. Two problems, local time shifting and spatial signal variation, should be solved to apply the time-warped similarity measure algorithms to wireless sensor networks. We find that our proposed algorithm can overcome those problems very efficiently and outperforms the other algorithms by at least 10.3% accuracy.

• Journal of Astronomy and Space Sciences
• /
• v.26 no.3
• /
• pp.287-294
• /
• 2009

In helioseismology it is normally assumed that p-mode oscillations are excited in a statistically independent fashion. Unfortunately, however, this issue is not clearly settled down in that two experiments exist, which apparently look in discrepancy. That is, Appourchaux et al. (2000) looked at bin-to-bin correlation and found no evidence that the assumption is invalid. On the other hand, Roth (2001) reported that p-mode pairs with nearby frequencies tend to be anti-correlated, possibly by a mode-coupling effect. This work is motivated by an idea that one may test if there exists an excess of anticorrelated power variations of pairs of solar p-modes. We have analyzed a 72-day MDI spherical-harmonic time series to examine temporal variations of p-mode power and their correlation. The power variation is computed by a running-window method after the previous study by Roth (2001), and then distribution function of power correlation between mode pairs is produced. We have confirmed Roth's result that there is an excess of anti-correlated p-mode pairs with nearby frequencies. On the other hand, the amount of excess was somewhat smaller than the previous study. Moreover, the distribution function does not exhibit significant change when we paired modes with non-nearby frequencies, implying that the excess is not due to mode coupling. We conclude that the origin of this excess of anticorrelations may not be a solar physical process, by pointing out the possibility of statistical bias playing the central role in producing the excess.

• Proceedings of the Korean Society of Marine Engineers Conference
• /
• 2005.11a
• /
• pp.220-221
• /
• 2005

Sub-bottom profiler(SBP) has been used extensively for the mapping of basement in the foundation design of offshore structure, for pre- and post-dredging operations within harbors and channels, for selection of pipeline routes, sitting of drilling platforms, and in the exploration for an aggregates such as sands and gravels. During the construction of Siwha embankment for irrigation water and the expansion of arable land, the breaking of an embankment unfortunately occurred so that a lot of riprap was swept away and widely dispersed by the tide and strong current. The feasibility study for the construction of the tidal-powered electric plant in Siwha embankment was performed quite recently. Therefore we made use of SBP survey to investigate the distribution of the lost riprap. We could successfully map out the distribution of the lost riprap from the reflection amplitude characteristics of the sediments in SBP data set. We demonstrated the variation of reflection amplitude versus the sediments with and/or without riprap by means of the numerical modeling of acoustic wave equation using finite difference method. Also we examined an amplitude anomaly of the ripraped area through the physical modeling using ultrasonic.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.3 no.1
• /
• pp.28-37
• /
• 2004

An on-line monitoring system of endmill failure such as weal, chipping, and fracture is developed using AE, cutting force Characteristic variations of AE and cutting force signals due to endmill failure are identified as follows. When endmill fracture occurs, AE count rate shows a rapid Increase in conjunction with a subsequent decrease while a standard deviation of the principal cutting force Increases significantly. The increase of AE count rate precedes the Increase of standard deviation of principal cutting force. Chipping results in relatively small increase and decrease of AE count rate without any significant variation of the cutting force Gradual increase of AE count rate and mean principal cutting force are Identified to be related with the wear of cutter. A cutter fracture detection algorithm is developed based on the present results. The signals me normalized to enhance the applicability of the algorithm to Wide those of fresh cutters, and qualitative characteristics of AE signals encountered at the moment of fracture are employed. It is demonstrated that the algorithm can detect the cutter fracture successfully.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.395-398
• /
• 1999

본 연구에서는 초음파현미경에서 다중의 동작주파스를 사용하여 얻어진 영상을 합성하여 깊이분해능이 개선된 초음파영상을 복원하는 방법을 연구하였다. 실험을 위하여 중심주파수가 5MHz이고 비대역폭이 $35\%$인 초음파변환기를 사용하여 초음파현미경을 구성하였다. 시편은 알루미늄을 이용하여 표면에 깊이가 $100{\mu}m$이고 지름이 2mm인 원형결함을 제작하였다. 실험결과, 동작주파수가 5MHz인 단일 주파수를 사용하여 얻어진 영상에서 결함이 존재하는 면과 그렇지 않은 면에 대한 콘트라스트가 0.098로 나타났다. 반면에 동작주파수를 $4.4MHz\~5.6MHz$ 범위에서 200kHz씩 변화시키면서 획득한 데이터를 합성하여 얻어진 영상에서 결함이 존재하는 면과 그러하지 않은 면에 대한 영상 콘트라스트가 0.398로 나타났다. 따라서 초음파현미경에서 다중의 동작주파수를 사용한 경우, 피사체의 깊이 변화에 대한 영상의 콘트라스트가 단일주파수를 사용한 경우에 비하여 개선되어 나타났으며 깊이분해능이 향상되었다. 앞으로의 과제는 초음파현미경에서 음향렌즈의 집속도 변화에 대한 적절한 동작주파수의 동작범위와 영상의 분해능과의 관계에 대한 연구가 계속되어져야 한다고 본다.