• 대한음성학회지:말소리
• /
• 제59호
• /
• pp.53-68
• /
• 2006

We report the evaluation results of the Korean speech recognition platform called ECHOS. The platform has an object-oriented and reusable architecture so that researchers can easily evaluate their own algorithms. The platform has all intrinsic modules to build a large vocabulary speech recognizer: Noise reduction, end-point detection, feature extraction, hidden Markov model (HMM)-based acoustic modeling, cross-word modeling, n-gram language modeling, n-best search, word graph generation, and Korean-specific language processing. The platform supports both lexical search trees and finite-state networks. It performs word-dependent n-best search with bigram in the forward search stage, and rescores the lattice with trigram in the backward stage. In an 8000-word continuous speech recognition task, the platform with a lexical tree increases 40% of word errors but decreases 50% of recognition time compared to the HTK platform with flat lexicon. ECHOS reduces 40% of recognition errors through incorporation of cross-word modeling. With the number of Gaussian mixtures increasing to 16, it yields word accuracy comparable to the previous lexical tree-based platform, Julius.

• 한국군사과학기술학회지
• /
• 제12권1호
• /
• pp.8-17
• /
• 2009

This paper proposes robust ship wake search method in the target evasion environment. Moving surface ships generate a long trailing wake in the rear of a surface ship. Wake homing torpedo sensing this wake can detect the surface target and engage it automatically. In wake homing torpedo, wake search method is important element to maximize effectiveness of wake homing torpedo. This paper proposes one-side, two-side and centering mode according to passing wake boundary scenarios. Also, wake deflection angle is deduced by using the principle of deflection angle of acoustic torpedo. The representative experimental results using monte-carlo simulation demonstrate that the searching method using one-side mode is superior to two-side and centering mode in the target evasion environment.

• Journal of Advanced Marine Engineering and Technology
• /
• 제34권8호
• /
• pp.1222-1229
• /
• 2010

비선형 측심기를 이용한 초고해상 음향탐사는 기존의 고해상 음향탐사보다 투과력은 작지만 해상력이 뛰어나기 때문에 천해역에서의 해저환경조사에 유용하게 사용된다. 비선형 지층탐사 시스템은 정확한 수심측정 뿐 아니라 퇴적층과 하부구조에 대한 상세한 정보를 제공한다. 특히 상부 퇴적층의 정밀층후 및 퇴적구조 확인을 통해 해저에 파묻혀 있는 pipeline, 침몰선박, 인위적 물체 등에 대한 탐색에 이용된다. 이 연구에서는 초고해상 지층탐사기를 이용하여 얻어진 지층탐사 기록과 지질자료를 대비하여 오염퇴적층을 구분하였다. 그리고 지층 기록에 나타나는 음향이상을 해석하여 해저에 매설된 케이블의 매설 깊이를 검측하였다.

• 한국지구물리탐사학회:학술대회논문집
• /
• 한국지구물리탐사학회 2008년도 공동학술대회
• /
• pp.79-84
• /
• 2008

초고해상 음향탐사는 기존의 고해상 음향탐사보다 투과력은 작지만 해상력이 뛰어나기 때문에 천해역에서의 해저환경조사에 유용하게 사용된다. 특히 상부 퇴적층의 정밀 층후 및 퇴적구조 확인을 통해 해저에 파묻혀 있는 pipeline, 침몰선박, 인위적 물체 등에 대한 탐색에 이용된다. 이 연구에서는 초고해상 지층탐사기를 이용하여 얻어진 지층탐사 기록과 지질자료를 대비하여 오염퇴적층을 구분하였다. 그리고 지층 기록에 나타나는 음향이상을 해석하여 해저에 매설된 케이블의 매설 깊이를 검측하였다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1999년도 제14차 학술회의논문집
• /
• pp.149-152
• /
• 1999

The aim of the work described in this paper is to develop a complex underground acoustic system which detects and localizes the origin of an underground hammering sound using an array of hydrophones located about loom underground. Three different methods for the sound localization will be presented, a time-delay method, a power-attenuation method and a hybrid method. In the time-delay method, the cross correlation of the signals received from the way of sensors is used to calculate the time delays between those signals. In the power-attenuation method, the powers of the received signals provide a measure of the distances of the source from the sensors. In the hybrid method, both informations of time-delays and power-ratios are coupled together to produce better performance of position estimation. A new acoustic imaging technique has been developed for improving the hybrid method. For each method the sound localization is carried out in three dimensions underground. The minimum distance between the true and estimated origins of the source is 28 m for a search area of radius 250m.

• 로봇학회논문지
• /
• 제13권2호
• /
• pp.79-85
• /
• 2018

This study focuses on autonomous exploration based on map expansion for an underwater robot equipped with acoustic sonars. Map expansion is applicable to large-area mapping, but it may affect localization accuracy. Thus, as the key contribution of this paper, we propose a method for underwater autonomous exploration wherein the robot determines the trade-off between map expansion ratio and position accuracy, selects which of the two has higher priority, and then moves to a mission step. An occupancy grid map is synthesized by utilizing the measurements of an acoustic range sonar that determines the probability of occupancy. This information is then used to determine a path to the frontier, which becomes the new search point. During area searching and map building, the robot revisits artificial landmarks to improve its position accuracy as based on imaging sonar-based recognition and EKF-SLAM if the position accuracy is above the predetermined threshold. Additionally, real-time experiments were conducted by using an underwater robot, yShark, to validate the proposed method, and the analysis of the results is discussed herein.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2004년도 추계학술대회논문집
• /
• pp.810-813
• /
• 2004

The task of introducing sounds leads to the realization of a comfortable acoustic environment and the design of good soundscape. This paper aims to rate the preference of environmental sounds suitable to the public spaces such as park, garden, bus terminal, urban street. And aims to search for the methodology how to select the introducing sounds by subjective and objective procedures. Seventeen kinds of introducing sounds were rated by nine adjectives in the presence of visual site informations. Also adequate sound levels were calculated by adjusting the volume of introducing sounds in the presence of the background sounds and visual informations. The results of psycho-acoustic experiments are as follows. 1. Proposal of various introducing sounds were evaluated to be possible at the rather calm natural park. 2. The introducing sounds which have opposite visual image or already existing at the place were evaluated not adequate to the sound circumstances of the place. 3. By the sounds with the same meaning, the identity of the street and the improvement of sound circumstance can be obtained.

• ETRI Journal
• /
• 제44권3호
• /
• pp.462-475
• /
• 2022

Although researchers have proposed numerous techniques for speech emotion recognition, its performance remains unsatisfactory in many application scenarios. In this study, we propose a speech emotion recognition model based on a genetic algorithm (GA)-decision tree (DT) fusion of deep and acoustic features. To more comprehensively express speech emotional information, first, frame-level deep and acoustic features are extracted from a speech signal. Next, five kinds of statistic variables of these features are calculated to obtain utterance-level features. The Fisher feature selection criterion is employed to select high-performance features, removing redundant information. In the feature fusion stage, the GA is is used to adaptively search for the best feature fusion weight. Finally, using the fused feature, the proposed speech emotion recognition model based on a DT support vector machine model is realized. Experimental results on the Berlin speech emotion database and the Chinese emotion speech database indicate that the proposed model outperforms an average weight fusion method.

• 한국군사과학기술학회지
• /
• 제27권3호
• /
• pp.416-427
• /
• 2024

The wake-homing underwater guided weapon that detects and tracks wake generated during voyage of a surface ship is impossible to avoid with the present acoustic deception torpedo defense system. Therefore, research on bubble-generating wake decoy is necessary to deceive wake-homing underwater guided weapon. Experiments in various environments are required to verify the effective operation method and performance of the wake decoy, but performance verification through underwater experiment is limited. In this paper, we develop a simulator for an torpedo defense system of surface ship, which is applied bubble-generating wake decoy, against acoustic, wake, and hybrid homing underwater guided weapon attack. The simulator includes surface ship model, acoustic decoy(static, mobile) model, bubble-generating wake decoy model, search and motion model of underwater guided weapon and so on. By integrating various models, MATLAB GUI simulator was developed. Through the simulation results for various environmental variables by this simulator, it is judged that effective operation method and performance verification of the bubble-generating wake decoy can be performed.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 1998년도 The Third Asian Fuzzy Systems Symposium
• /
• pp.495-500
• /
• 1998

This paper introduces our three approaches to reduce the burden of human interactive EC operators: (1) improvement of the interface of presenting individuals, (2)improvement of the interface of inputting fitness values, and (3) fast EC convergence. We propose methods to display individuals in order of predicted fitness values by neural networks or Euclidean distance measure for (1), to input quantized fitness values for (2), and to make a new elite by approximating the EC search space with a quadratic function for (3). They are evaluated through simulations and subjective testes, and their effects have shown.