• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 1998년도 제10회 한글 및 한국어 정보처리 학술대회
• /
• pp.159-163
• /
• 1998

다국어 대화체 번역 시스템은 미국의 카네기 멜론 대학과 일본의 ATR 및 한국의 전자통신연구원 등이 가입한 CSTAR의 99년 국제간 음성언어번역 시스템 데모를 위한 한국어측 번역 시스템이다. CSTAR-IF는 국제간 데모를 위해 각 국의 시스템끼리 주고 받는 정보의 단위 혹은 형태로서, 중간언어 표현의 한 가지 방법으로 간단하면서도 단순한 표현으로 특정 영역 내에 나타나는 의미를 표현할 수 있도록 정의되었다. 다국어 번역 시스템은 크게 두 가지로 나누어진다. 하나는 한국어 음성인식 결과를 IF로 변환하는 해석 시스템이고, 다른 하나는 IF로부터 한국어 문장을 생성하여 음성으로 들려주는 생성 시스템이다. 한국어 해석 시스템은 현재 92%의 해석 성공률을, 생성 시스템은 98%의 생성 성공률을 보이고 있다.

• 음성과학
• /
• 제15권4호
• /
• pp.29-41
• /
• 2008

The purpose of this paper is to examine how Chinese speakers realize Korean prosodic phrases in the reading of Korean texts. Prosodic phrase, in this study, is defined as basic unit of spoken language which can be perceived as purely separate phonetic unit by both hearer and speaker, and is realized with a coherent intonational configuration. Prosodic phrase plays an important role in both speech production and perception. In the second language acquisition, prosody influences the accuracy and fluency of spoken language. The main purpose of this study is to describe the aspect of syntagmatic operation of prosody that produces prosodic phrases. We have specifically examined the relations between the prosodic phrase's boundary and its syntactic status. Furthermore, we examined internal syntactic structure of each prosodic phrase. And the results of each analysis were compared to the aspects of prosodic phrases' formation produced by native Korean speakers. The results show that Chinese speakers tend to coincide the prosodic phrases with syntactic structure more than native Korean speakers.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2016년도 제28회 한글 및 한국어 정보처리 학술대회
• /
• pp.217-219
• /
• 2016

본 논문은 형태소 분석 및 품사 태깅을 위해 seq2seq 주의집중 모델을 이용하는 접근 방법에 대하여 기술한다. seq2seq 모델은 인코더와 디코더로 분할되어 있고, 일반적으로 RNN(recurrent neural network)를 기반으로 한다. 형태소 분석 및 품사 태깅을 위해 seq2seq 모델의 학습 단계에서 음절 시퀀스는 인코더의 입력으로, 각 음절에 해당하는 품사 태깅 시퀀스는 디코더의 출력으로 사용된다. 여기서 음절 시퀀스와 품사 태깅 시퀀스의 대응관계는 주의집중(attention) 모델을 통해 접근하게 된다. 본 연구는 사전 정보나 자질 정보와 같은 추가적 리소스를 배제한 end-to-end 접근 방법의 실험 결과를 제시한다. 또한, 디코딩 단계에서 빔(beam) 서치와 같은 추가적 프로세스를 배제하는 접근 방법을 취한다.

• 한국음향학회:학술대회논문집
• /
• 한국음향학회 2002년도 하계학술발표대회 논문집 제21권 1호
• /
• pp.33-36
• /
• 2002

We implemented a robust Korean broadcast news transcription system for out-of-vocabulary (OOV), tested its performance. The occurrence of OOV words in the input speech is inevitable in large vocabulary continuous speech recognition (LVCSR). The known vocabulary will never be complete due to the existence of for instance neologisms, proper names, and compounds in some languages. The fixed vocabulary and language model of LVCSR system directly face with these OOV words. Therefore our Broadcast news recognition system has an offline OOV update module of language model and vocabulary to solve OOV problem and selects morpheme-based recognition unit (so called, pseudo-morpheme) for OOV robustness.

• 한국어교육
• /
• 제28권4호
• /
• pp.1-38
• /
• 2017

This longitudinal study analyzed syntactic development (Complexity, Accuracy, and Fluency; CAF) of six L2 Korean learners in a natural learning context. The learners recalled the stories of a short animated video through speaking and writing every 3 months, from month 0 to 15. The learners' responses were analyzed for a series of CAF measures and their cognitive, psychological, and social variables were investigated. The results showed that (i) L2 Korean learners' speaking and writing in various time periods showed significant differences in spoken and written accuracy, and complexity; (ii) the correlation between spoken and written complexity, spoken and written accuracy, as well as spoken and written fluency were significant, and (iii) the regression analysis showed that learners' cognitive, social, and psychological variables have significant effect on the L2 Korean syntactic development. The current study reports that L2 Korean learners engaged in self-learning in a natural learning environment without formal instruction made significant syntactic development.

• 한국언어정보학회:학술대회논문집
• /
• 한국언어정보학회 1992년도 Language, Information and Computation = Proceedings of Asian Conference, Seoul
• /
• pp.1-13
• /
• 1992

• 한국언어정보학회:학술대회논문집
• /
• 한국언어정보학회 1995년도 Language, Information and Computation = Proceedings of the 10th Pacific Asia Conference, Hong Kong
• /
• pp.231-236
• /
• 1995

• The Journal of the Acoustical Society of Korea
• /
• 제17권2E호
• /
• pp.38-46
• /
• 1998

In this paper, we propose a remote speech input device, a new method of user-friendly speech input in spontaneous speech recognition system. We focus the user friendliness on hands-free and microphone independence in speech recognition applications. Our method adopts two algorithms, the automatic speech detection and the microphone array delay-and-sum beamforming (DSBF)-based speech enhancement. The automatic speech detection algorithm is composed of two stages; the detection of speech and nonspeech using the pitch information for the detected speech portion candidate. The DSBF algorithm adopts the time domain cross-correlation method as its time delay estimation. In the performance evaluation, the speech detection algorithm shows within-200 ms start point accuracy of 93%, 99% under 15dB, 20dB, and 25dB signal-to-noise ratio (SNR) environments, respectively and those for the end point are 72%, 89%, and 93% for the corresponding environments, respectively. The classification of speech and nonspeech for the start point detected region of input signal is performed by the pitch information-base method. The percentages of correct classification for speech and nonspeech input are 99% and 90%, respectively. The eight microphone array-based speech enhancement using the DSBF algorithm shows the maximum SNR gaing of 6dB over a single microphone and the error reductin of more than 15% in the spontaneous speech recognition domain.

• 한국정보처리학회논문지
• /
• 제6권8호
• /
• pp.2113-2123
• /
• 1999

부분 자유어순 특성을 가지는 한국어를 CFG형태의 문법으로 기술했을 때 문법이 방대해지고 CFG형태의 문법을 파서가 이용할 때는 자연발화문의 특징인 간투어, 중복발화 등 때문에 불필요 성분을 따로 처리해야 하므로 파서의 부담이 커진다. 이러한 문제점을 해결하기 위해 본 논문에서는 발화문에서 중요한 의미를 가지는 요소를 '핵심개념'이라 정의하고 핵심개념만을 문법에 기술하여 문법이 방대해지는 것을 막고 문법에 기술된 핵심개념을 파싱요소로 선택함으로써 불필요 성분처리에 대한 파서의 부담을 줄였으며 이렇게 단순화된 문법만으로도 정확한 파싱결과를 내줄 수 있음으로 보인다. 실험결과 '여행안내'영역 자연발화문에 대해서 평균 98%이상의 올바른 파싱결과를 얻어낼 수 있었다.

• 영어어문교육
• /
• 제9권2호
• /
• pp.133-149
• /
• 2003

As the world becomes a global village, national conferences become more frequent, and the language of choice to mediate the exchange of information is English. Accordingly, the purpose of this research is to find several methods to prepare L2 students for English communication at conferences. Various studies about discourse markers can be found in this study. Discourse markers are indicators of fluency in spoken language, and indicators of improvement from features of spoken English to more typical forms of formal writing. In this study the presentation styles of native speakers and Asian non-native speakers are compared, and the difference between efficient and non-efficient speakers at conferences are discussed. As a study of conference English teaching and learning method, this research targets suggestion and application of discourse markers for L2 speakers. As a result, the fact that discourse markers are very important in effective writing and conference English is found and accordingly we can see that developing the ability to use discourse markers in writing and speaking are essential for L2 speakers.