• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2002년도 11월 학술대회지
• /
• pp.47-50
• /
• 2002

This paper is to introduce 'Dr. Speaking', which was recently developed by Eonon Inc.. 'Dr. Speaking' is an English pronunciation tutoring system. This has three distinguishing features. First, it teaches how to organize a speaker's vocal organs to pronounce accurately. Second, after it compares a speaker's pronunciation with that of a native speaker's, it grades that speaker's pronunciation level according to phonetic standards. Third, it provides proper information necessary for correcting a speaker's incorrect pronunciation. It is not always easy for a tutoring system to execute the above three almost simutaneously. However, 'Dr. Speaking' proved itself that it is possible by adding speech technology (e.g. speech recognition) to phonetic knowledge.

• 대한음성학회지:말소리
• /
• 제35_36호
• /
• pp.49-61
• /
• 1998

The study on the intonational characteristics is concentrated on the lowering of the pitch level that is described as declination and downstep. The Taegu dialect, which has phonological accentual system, has these phonetic characteristics in affirmative sentences or Yes-No Question sentences. But there is the opposite phenomenon in WH question sentences in Taegu dialect. When the accent of interrogative word in the sentence intial position is LHL, intonation pattern shows a continuous upward movement, indicating that intonation pattern of Taegu dialect is influenced by not only grammatical system but also accentual system.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2006년도 춘계 학술대회 발표논문집
• /
• pp.37-40
• /
• 2006

In this paper, we propose a VQ codebook design of speech recognition feature parameters in order to improve the performance of a distributed speech recognition system. For the context-dependent HMMs, a VQ codebook should be correlated with phonetic distributions in the training data for HMMs. Thus, we focus on a selection method of training data based on phonetic distribution instead of using all the training data for an efficient VQ codebook design. From the speech recognition experiments using the Aurora 4 database, the distributed speech recognition system employing a VQ codebook designed by the proposed method reduced the word error rate (WER) by 10% when compared with that using a VQ codebook trained with the whole training data.

• 한국음향학회지
• /
• 제23권2호
• /
• pp.173-179
• /
• 2004

음소 질의어 집합은 문맥 속에서 비슷한 조음 효과를 보이는 음소들을 분류해 놓은 것으로서, 음성 인식 시스템 학습 시 결정트리를 기반으로 HMM (hidden Markov model)의 상태들을 클러스터링할 때 사용된다. 현재까지의 음소 질의어 집합은 대부분 음성학자나 언어학자들에 의해 수작업으로 제시되어 왔는데, 이러한 지식 기반음소 질의어들은 언어 또는 유사음소 단위 (PLU: phone like unit)에 종속될 뿐 아니라 생성된 클러스터 내의 동질성을 저하시킬 수 있다는 단점이 있다. 본 논문에서는 이와 같은 문제점들을 해결하기 위해 음성 데이터를 사용하여 측정한 음소들 사이의 유사도를 기반으로 언어나 유사음소단위에 상관없이 자동으로 음소 질의어 집합을 생성하는 알고리즘을 제안한다. 실험결과, 제안한 방법으로 생성된 음소 질의어들을 사용한 인식기의 에러율이 약 14.3%감소하여 데이터 기반의 음소 질의어 집합이 상태 클러스터링에 효율적임을 관측하였다.

• 디자인학연구
• /
• 제13권1호
• /
• pp.103-110
• /
• 2000

한글은 동아시아 한자문화권 속에서 태어났다. 우리의 글자문화는 서양 글자 문화는 엄연히 다른 전통이 있었다. 동양에서 글자란 영물이자 주술이었으며, 존엄과 신비 그리고 제의가 함유된 경외의 대상이었다. 동양에서는 글씨 그 자체의 엄연한 가치를 인정하는 문화적 전통이 있었던 것이다. 이러한 글자관 세계속에서 완벽한 소리글자로서 한글은 태어났다. 그러나 한글의 특성은 뜻글자로서의 한자적 배경과 사뭇 다르며 독특하다. 한글은 소리글자임에도 불구하고 한자적 이미지의 고향을 강하게 지니고 있다. 이것은 라틴 알파벳이 그 진화의 오랜 여정 속에서 이미지의 고향을 잃어버린 것과 대조되는 부분이며, 소리글자로서의 한글이 이미지의 속성을 함께 가지고 있는 것은 한글의 특성이라 할 수 있다. 다시 말해 한글 형태는 조합 방식에서 한자와 같은 속성을 가지고 있으면서, 소리글자로서의 특성은 라틴 알파벳에 가깝다. 한글은 곧 시간적으로 한자 문화에 바탕을 둔 시각 이미지 글자의 배경으로서의 기억을 생생히 가지고 있고, 또 기능적으로는 매우 발달된 형태의 소리글자로서의 글자의 두 가지 속성을 모두 한몸에 지니고 있는 것이 특징이다. 이러한 한글의 특징은 이미지 시대에 살고 있는 우리에게 바로 우리 시각문화의 존재 의미를 일깨워주는 부분이라 하겠다. 곧 세계 글자문화 속에서 돋보이는 이러한 한글 타이포그라피의 정체성이야 말로 바로 우리 시각문화의 본질이 될 것이다.

• 말소리와 음성과학
• /
• 제11권4호
• /
• pp.35-43
• /
• 2019

This study examines the voice onset times (VOTs) of Korean stops produced by Korean learners of English with high language proficiency and long L2 exposure (i.e., Korean-English bilinguals) to assess whether the VOTs of their lax and aspirated stops are merging and, if so, which types of stops are being changed. Thirteen Korean speakers (six female and seven male) who had studied in the USA for more than three to ten years participated. The results show that the speakers in this study with long L2 exposure are participating in the VOT merger, in which VOTs for aspirated stops are reduced while those for lax stops are increased. In other words, change in VOT affects not only aspirated stops but also lax stops. The results indicate that L1 phonetic drift may not be primarily affected by the amount of L2 exposure, and language contact may not be the primary factor triggering a sound change in the Korean stop system. Further study is necessary focusing on the phonetic shift of the "lax" category because it may play a pivotal role in a tonogenetic-like sound change in present-day Korean.

• 대한음성학회지:말소리
• /
• 제41호
• /
• pp.31-47
• /
• 2001

The purpose of this study is to explain the cause of obsturents neutralization in syllable final position. Most of the previous phonological studies did not reflect phonetic reality sufficiently because of the limited use of the binary feature system. Using binary distinctive features, we can't explain the cause of neutralization. In order to explain the cause of neutralization, I use the multi-valued phonetic feature -[vocal tract aperture]. By [vocal tract aperture] I mean the distance between articulators in the hold stage. In this study, I claim that the cause of neutralization is assimilation to [vocal tract aperture] 0 degree. The neutralized sounds become aplosives, as a consequence of assimilation to [vocal tract aperture].

• 대한음성학회지:말소리
• /
• 제11_14호
• /
• pp.73-89
• /
• 1987

This paper is an attempt to propose a new model of intonation called 'Configurational Contour Model'. For our purpose two previous models are discussed : one is the system of tonetic-stress marks and the other is the theory of pitch levels. To overcome shortcomings which these approaches display, the new model is based on two fundamental characteristics of intonation: (1) that intonation is a suprasyllabic phenomenon; and (2) that intonation is basically a matter of configurations, i.e., of rising, falling and level. Accordingly, the new model can be relatable to relevant aspects of the phonetic representation of utterances and also can express all the tonal oppositions in a given language.

• 말소리와 음성과학
• /
• 제1권3호
• /
• pp.103-108
• /
• 2009

This paper proposes a method for building a transliteration dictionary, which is based on pronouncing information extracted from two kinds of existing dictionaries. Also, it proposes a method for transforming the pronouncing information into Korean translitered words. To express the pronouncing information, we define Phoman code system. In order to avoid phonetic estimation process of English words which is the most important problem, the proposed method uses the pronouncing information extracted from the existing dictionaries. Therefore, unlike previous approaches, the proposed method does not need any incomplete phonetic estimation process so that it can produce accurate transliteration results. The proposed method has been fully implemented.

• 대한음성학회지:말소리
• /
• 제52호
• /
• pp.111-120
• /
• 2004

This paper focuses on the DSP implementation of an HMM-based speech recognizer that can handle several hundred words of vocabulary size as well as speaker independency. First, we develop an HMM-based speech recognition system on the PC that operates on the frame basis with parallel processing of feature extraction and Viterbi decoding to make the processing delay as small as possible. Many techniques such as linear discriminant analysis, state-based Gaussian selection, and phonetic tied mixture model are employed for reduction of computational burden and memory size. The system is then properly optimized and compiled on the TMS320C6711 DSP for real-time operation. The implemented system uses 486kbytes of memory for data and acoustic models, and 24.5 kbytes for program code. Maximum required time of 29.2 ms for processing a frame of 32 ms of speech validates real-time operation of the implemented system.