페이징 채널 지연시간 분석

정용주 한국경영과학회 2000 한국경영과학회 학술대회논문집 Vol.- No.1

이동통신 시스템에서 가입자수가 증가하고 다양한 서비스가 구현됨으로써 전체적인 장비 또는 자원에 대한 부하가 증거하고 있다. 특히, 셀의 반경이 작아지고 단문(Short Message)을 활용한 다양한 부가 서비스가 개발됨으로써 이와 관련된 정보를 전달하는 페이징 메시지의 분류도 복잡해지고 페이징 채널의 부하도 커지고 있다. 본 연구에서는 먼저, 이동통신시스템에 페이징 채널의 구조와 역할을 분석하고 페이징 채널에서 전송하는 메시지를 그 기능과 특성에 따라 분류한다. 긴급하게 전송되어야 하는 메시지를 보호하기 위한 우선권 부여 방법을 모색하고, 각 방법에 있어서 메시지 종류별 전송지연시간(Delay)을 수리적으로 도출하고 결과를 비교한다.

영어음의 지각

정용주 진주교육대학교 1989 論文集 Vol.33 No.1

The purpose of this paper is to consider the perception of language, the first step of comprehension process, through literature of psycholinguistics. When speech sound is seen through spectrogram, it produces the blending segments, not a series of descrete ones. Speech sounds are continuous stream which cannot he segmented artificially and do not have invariance that sounds the same in all contexts. Phonetic cued available in the phonetic stage are context- independent and context - dependent cues. These two cues are used in identifying consonante. One of the important context - dependent cues are voice onset time. Context - dependent curia themselves vary as the context of phoneme changes. This means that many phonemes do not have invariant properties. When we want to identify a particular phoneme we must identify the ones preceding and following it. This phenomenon conveying information of more than one phoneme at a time is called parallel transmission. Vowels are characterized in terms of unvarying, steady- state formant patterns. Vowels are identified by the difference of formant frequencies. Hut vowels maintaining steady - state formant are affected by the context in which they occur. Consequently vowels are identified by the changing formant patterns of the phonemes that precede and fellow them. Many phonetic cues are carried by the formant patterns and by the changing way of formants over time. Step consonants like /b/, /d/, and /g/ are identified by the changes of formants called formant transition. Fricative and nasal consonants also involve formant transition. But fricative consonants hold distinctive noise lacking formant structure, and nasal consonants have the distinctive nasal formants. Even though speech signal varies a lot, its variance does not greatly affect a listener trying to identify it. Generally listners do not perceive the variance of speech signal. Only when a listener identify the signal as a different phoneme, he listens differently. This phenomenon is referred as categorical perception. After Liberman and his colleague first reported categorical characteristics of language perception, many experiments have showed that speech sound is perceived categorically. It is kown that even infants perceive stop consonants categorically. Perception process of speech stimuli seems to involve a system of feature detectors. This is mechanisms that respond in all-or-none manner to the presence or absence of a particular properly in the speech signal. The evidence for the existence of feature detectors is mainly indirect. Neurological evidence for such detectors is not yet obtained. When speech signals do not have enough information, we can not identify phonological features unambiguously. This means that through the phonological analysis of the speech signal itself speech perception is impossible. When listeners analyze the speech signal, they use the information of lexical, syntactic, and semantic properties. Thus the phonological analysis of the speech signal involves the synthesis of many informations. There is still al great deal to be learned about active, synthetic aspect of speech perception.

HMM 분할정보를 이용한 Stochastic Segment 모델의 분별적 학습에 대한 연구

정용주 啓明大學校 産業技術硏究所 1999 産業技術硏究所 論文報告集 Vol.22 No.2

In this paper, we propose a discriminative training algorithm for the stochastic segment model(SSM). As the SSM is usually trained by maximum likelihood estimation (MLE), a discriminative training algorithm is required to improve the recognition performance. To reduce the computational complexity and storage amount in the discriminative training of SSM, a hybrid architecture of SSM and HMM is proposed. With the discriminative training of the SSM, the word error rate is reduced by 17% compared with the MLE-trained SSM in the speaker-independent speech recognition.

IMT-2000 시스템의 데이터 트래픽 용량산출을 위한 시뮬레이터

정용주 부산 외국어 대학교 2004 外大論叢 Vol.29 No.-

This aim of this study is to develop a simulator, which is used to identify the performance in view of data traffic of mobile communication networks, especially of cdma2000-1X networks, considering the dynamics of data traffics and the process of data transmission such as packet scheduling. We present a simulator fully incorporating packet handling process of cdma2000-1X data network as well as three-level ON/OFF traffic model describing the behavior of source data traffics.

다양한 지르코늄 열전달 표면에서의 니켈 부착

정용주,연제원,최광순,최계천,김원호 한국방사성폐기물학회 2005 한국방사성폐기물학회 학술논문요약집 Vol.3 No.1

언어습득과정과 이해력의 발달

정용주 진주교육대학교 1986 論文集 Vol.30 No.1

Childern begin to produce recognizable utterances at abort the age of one, and by the age of four or five their utterances become nearly like those of adults. This development that occurs with rapidity and relatively no effort poses an important problem on the understanding of language development. General views of language acquisition emphasized the child-adult interactions, that is, reinforcement, imitation, expansion, and parental input. Studies on the effects of reinforcement indicate that it could not have a major influence on child language development. Adult's verbal reinforcement of the child's utterances depends on truth value rather then syntactic well-formedness. Therefore, child could not learn adult's syntax through reinforcement. The effects of child's imitation of adult utterances are less clearcut. Some studies find that children's imitative utterances are less advanced than their other utterances, but one study indicates that some particular words or syntactic structures tend to occur earlier in the imitative utterances than the nonimitative ones. Steinberg suggests that imitation could facilitate the acquisition of speech through the direct copying of speech forms and through motivating the child to make rule adjustments. However, their is no evidence that the linguistic ability of children who imitate more grow faster than that of children who imitate less. Thus imitation is not likely to play a crucial role in the development of language, Expansion, defined as expanding a child's fragmentary utterance into complete, correct one, does not seem to be influential to language development. Children whose speech is frequently expanded show slightly greater progress than those whose speech is not expanded. However, its effect appears to be small. There is some evidence showing that syntactic properties of mother's speech play a role in the child language development. However, children do not grow in the world where specific situation constitutes only the form of specific utterance. Children should team that the different ways of expression mean the same communicative intent. Parental input becomes a basis on a child's forming hypothesis and provides materials of doing it. How childern use them depends on their linguistic, cognitive ability of a particular development level. The characteristics of language itself and semantic complexity are more influential to language development than the linguistic interactions between adults and children. Semantic complexity expressed by the syntactic structures determines how easily and fast a child begin to use the syntactic forms. Syntactic complexity is also a factor of language acquisition. Children's language development is a reflection of their cognitive development. Many linguists have argued that children learn language by formulating hypothesis about structures and meanings underlying utterances they hear. Children's hypothesis are not like the adult's grammar. They find errors and form a new hypothsis, more approximating to adult's grammar. This hypothesis-testing theory accounts for many data about language development, but does not account for all of them. Strategy development theory suggests that children selectively attend to some characteristics of utterances they hear. They develop strategies associating the cues of surface structure with the underlying structure. At the early stage children pay attention to the meanings of words and relations between them. Gradually they attend to the order in which the words occur, identifying the subject-verb-object structure as the basic word order. The next thing children will do for language development is to acquire additional strategies interpreting the exceptional structures to the general word order. Generally strategy development theory seems to account for the data of language acquisition better than hypothesis-testing theory. However, strategy development theory is far from satisfactory, and has much to be solved about the acquisition of child language.

방송뉴스 인식에서의 잡음 처리 기법에 대한 고찰

정용주,Chung Yong-joo 대한음성학회 2004 말소리 Vol.50 No.-

Recently, broadcast news speech recognition has become one of the most attractive research areas. If we can transcribe automatically the broadcast news and store their contents in the text form instead of the video or audio signal itself, it will be much easier for us to search for the multimedia databases to obtain what we need. However, the desirable speech signal in the broadcast news are usually affected by the interfering signals such as the background noise and/or the music. Also, the speech of the reporter who is speaking over the telephone or with the ill-conditioned microphone is severely distorted by the channel effect. The interfered or distorted speech may be the main reason for the poor performance in the broadcast news speech recognition. In this paper, we investigated some methods to cope with the problems and we could see some performance improvements in the noisy broadcast news speech recognition.

Bayesian 적응 방식을 이용한 잡음음성 인식에 관한 연구

정용주 한국음향학회 2001 韓國音響學會誌 Vol.20 No.2

An expectation-maximization (EM) based Bayesian adaptation method for the mean of noise is proposed for noise-robust speech recognition. In the algorithm, the on-line testing utterances are used for the unsupervised Bayesian adaptation and the prior distribution of the noise mean is estimated using the off-line training data. For the noisy speech modeling, the parallel model combination (PMC) method is employed. The proposed method has shown to be effective compared with the conventional PMC method for the speech recognition experiments in a car-noise condition. 본 논문에서는 잡음에 강인한 음성인식을 위해서 expectation-maximization (EM) 방식을 이용하여 잡음의 평균값을 추정하는 새로운 알고리듬을 제안하였다. 제안된 알고리듬에서는 온라인상의 인식용 음성이 직접 Bayesian 적응을 위해서 사용되며, 또한 훈련데이터를 이용하여 잡음의 평균값에 대한 사전 (prior) 분포를 알아낸 후 Bayesian 적응시에 이용한다. 잡음 음성의 모델링을 위해서는 PMC (parallel model combination) 방식을 이용하였고, 제안된 방식을 이용하여 자동차 잡음 환경 하에서 인식 실험을 수행한 결과, 기존의 PMC 방식에 비해서 향상된 인식성능을 보임을 알 수 있었다.

결정적 잡음 모델을 이용한 효율적인 잡음음성 인식 접근 방법

정용주 한국음향학회 2002 韓國音響學會誌 Vol.21 No.6

본 논문에서는 잡음음성 HMM (Hidden Markov Model)의 파라미터 값을 효율적으로 추정하는 새로운 방법에 대해서 제안하였다. 기존의 방법들에서 잡음음성의 HMM 파라미터 값을 추정하기 위해서는 먼저 잡음음성의 생성 모델을 가정한 후, 잡음과 원래 음성의 통계 모델을 이용하여 잡음음성 HMM 파라미터 값을 해석적으로 얻게 된다. 하지만 이러한 해석적 방법은 항상 단순화의 가정을 취하게 되므로 실제의 잡음음성 HMM 분포에 정확히 근접하는데 어려움을 겪게 된다. 본 연구에서는 이러한 가정을 하지 않고, 원래의 깨끗한 음성에서 얻을 수 있는 HMM의 파라미터 값을 사용하고 결정적 잡음 모델을 이용함으로서 기존의 방법보다 인식시에 계산량을 줄일 수 있었을 뿐만 아니라 인식 성능의 향상도 이룰 수 있었다. In this paper, we proposed an efficient method that estimates the HMM (Hidden Marke Model) parameters of the noisy speech. In previous methods, noisy speech HMM parameters are usually obtained by analytical methods using the assumed noise statistics. However, as they assume some simplication in the methods, it is difficult to come closely to the real statistics for the noisy speech. Instead of using the simplication, we used some useful statistics from the clean speech HMMs and employed the deterministic noise model. We could find that the new scheme showed improved results with reduced computation cost.

An Efficient Model Parameter Compensation Method for Robust Speech Recognition

정용주 대한음성학회 2003 말소리 Vol.45 No.-