• 한국음향학회:학술대회논문집
• /
• 한국음향학회 1998년도 제15회 음성통신 및 신호처리 워크샵(KSCSP 98 15권1호)
• /
• pp.81-86
• /
• 1998

트리 기반 모델링 기법 중 하나인 CART 방법을 이용하여, 운율구 추출과 운율구 사이의 휴지 기간을 모델링 하고자 한다. 모델링을 위한 특징 변수들의 유효성을 실험에 앞서 알아본 후, 생성된 트리들을 해석함으로써 제안하는 특징 변수들이 효과적임을 보인다. 음성 정보를 제외한 문서 정보만을 이용하여 실험한 결과, 운율구 경계 결정 오류율은 14.46% 이었고, 휴지 기간 예측 RMSE 가 132.61 msec 이었다.

• 한국음향학회지
• /
• 제16권1호
• /
• pp.24-32
• /
• 1997

이 논문에서는 첫 번째로 운율구 경계를 자동으로 추출하고자 할 때 인간의 지각과 관련하여 어떠한 자질을 이용하는 것이 좋은가 하는 점을 밝혀 보았다. 운율구 경계의 유형은 크게 경계가 없는 강도(zero), 약한 경계 강도(minor break strength), 강한 경계 강도(major break strength) 3 단계로 정하는 것이 통계적으로 의의가 있으며 지속시간의 정보가 강한 경계 유형의 결정에 중요하게 작용하는 자질이었음을 알 수 있었다. 두 번째로는 운율구 경계현상 분석결과를 바탕으로 운율구 경계의 경계 강도를 약한 경계 강도(zero를 포함)와 강한 경계 강도로 나누고, 2단계의 경계 강도를 텍스트상에서 문장성분의 bigram, trigram을 정보를 이용하여 자동으로 할당하였다. Bigram의 경우 Test-I, Test-II 텍스트 DB에 대해 각각 46.0%, 38.2%의 강한 경계 강도 예측정확률과 22.8%, 8.4%의 삽입오류율의 성능을 내었으며, Trigram인 경우 Test-I, Test-II 텍스트 DB 각각에 대해 58.3%, 42.8%의 강한 경계 강도 예측정확률과 30.0%, 11.8%의 삽입오류율를 나타냈다.

• 대한음성학회지:말소리
• /
• 제39호
• /
• pp.15-24
• /
• 2000

This study aims to reexamine Korean Lenis Stop Voicing (henceforth, LSV) and to specify its phonetic conditions in phonetic terms. LSV optionally occurs within certain prosodic domains. They are called 'Malthomak'(Lee, 1996),'phonological phrase'(Kang, 1992), or 'accentual phrase'(Jun, 1993). On the basis of Jun's phrasing, this study focuses on the more specific phonetic conditions of LSV in the accentual phrase medial position, sub-classifying voicing as complete and partial. The results shows that whether the stops become completely voiced or partially voiced was determined by the various phonetic environments, such as adjacent segments and following intonational phrase boundaries. It is shown that the conditions of LSV should be described in terms of more detailed phonetic environments and that they could be used in predicting the class of voicing.

• 음성과학
• /
• 제4권1호
• /
• pp.35-46
• /
• 1998

In this paper production and perception data were examined to discover what acoustic features are used in distinguishing wh-questions and yes/no-questions. Production data show that the two question types are distinguished by different accentual phrasing, pitch ranges in wh-phrases, and initial lenis stop voicing of the first syllable in verb phrases. Perception data by synthetic intonation show that the two question types are distinguished by the width of pitch ranges between the first and the second syllable in wh-phrases. Initial lenis stop voicing of the first syllable in verb phrases produces a strong effect on the perceptual discrimination of the two question types.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2007년도 한국음성과학회 공동학술대회 발표논문집
• /
• pp.114-118
• /
• 2007

The current study investigates the degree to which various prosodic cues at the boundaries of a prosodic phrase in Korean (Accentual Phrase) contributed to word segmentation. Since most phonological words in Korean are produced as one AP, it was hypothesized that the detection of acoustic cues at AP boundaries would facilitate word segmentation. The prosodic characteristics of Korean APs include initial strengthening at the beginning of the phrase and pitch rise and final lengthening at the end. A perception experiment revealed that the cues that conform to the above-mentioned prosodic characteristics of Korean facilitated listeners' word segmentation. Results also showed that duration and amplitude cues were more helpful in segmentation than pitch. Further, the results showed that a pitch cue that did not conform to the Korean AP interfered with segmentation.

• 한국음향학회지
• /
• 제17권6호
• /
• pp.43-53
• /
• 1998

본 논문에서는 한국어 TTS시스템을 위한 운율구 추출, 운율구 사이의 휴지 기간, 음소의 지속 시간 모델링 방법을 설명한다. 실험을 위해 여러 장르로 구성된 400문장을 선 정하고, 이를 전문 여성 아나운서가 발성하였다. 녹음된 음성 신호에 대해 음소 및 운율구 경계를 결정하고, 문장에 대해서는 형태소 분석, 발음표기 변환, 구문 분석을 수행하였다. 400문장(약33분) 중 240문장(약20분)을 이용하여 결정 트리 및 회귀 트리를 학습시킨 후, 160분장(약13분)에 대해 실험하였다. 운율 모델링을 위한 특징들이 제안되었고, 학습된 트리 들을 해석함으로써 특징들의 유효성이 평가되었다. 실험 문장에 대해 운율구 경계의 유무를 결정하는 결정 트리의 오류율은 14.46%이었고, 운율구 사이의 휴지 기간과 음소 지속 시간 을 예측하기 위한 회귀 트리들의 평균 제곱 오류근(RMSE)이 각각 132msec, 22msec이었다. 수집된 모든 자료(400문장)로 학습한 결과, 운율구 경계 결정 오류율, 휴지 기간 및 지속시 간 RMSE의 10-fold cross-validation 추정치가 각각 13.77%, 127.91msec, 21.54msec이었다.

• 음성과학
• /
• 제2권
• /
• pp.95-108
• /
• 1997

The purpose of this experimental study was to investigate prosodic characteristics of English as spoken by Koreans. Test materials were four English words, a sentence, and a paragraph. Six female Korean speakers and five native English speakers participated in acoustic and perceptual experiments. Pitch and duration of word syllables were measured from signals and spectrograms made by the Signalize 3.04 software program for Power Mac 7200. In the perceptual experiment, accent position, intonation patterns, rhythm patterns and phrasing were evaluated by the five native English speakers. Preliminary results from this limited study show that prosodic characteristics of Koreans include (1) pitch on the first part of a word and sentence is lower than that of English speakers, but the pitch on the last part is the opposite; (2) word prosody is quite similar to that of an English speaker, but sentence prosody is quite different; (3) the weakest point of sentence prosody spoken by Koreans is in the rhythmic pattern.

• 말소리와 음성과학
• /
• 제3권1호
• /
• pp.67-75
• /
• 2011

In Korean, 'Aspiration Merger' is the result of the heteromorphemic sequence of lenis stop and /h/ becoming a single aspirated stop word-medially. However, the contrast between lenis stop-plus-/h/ and an underlying aspirated stop is maintained when they span Phonological Phrase boundaries. By varying the position in the prosodic domain such as APP (Across Phonological Phrase) and PPM (Phonological Phrase Medial) positions, the phonetic properties of the two categories are compared. In the results from noise duration and change of intensity, lenis stop-plus-/h/ show a large difference between the APP and PPM positions. The results from a noise duration comparison show that the two categories are completely neutralized into aspirated stop in the PPM position and the complete neutralization is sensitive to prosodic phrasing.

• 대한음성학회지:말소리
• /
• 제35_36호
• /
• pp.145-155
• /
• 1998

This study intended to determine the most appropriate POS(Part of Speech) sets for predicting prosodic boundary strength efficiently. We used 3-level POB bets which Kim(1997), one of the authors, has devised. Three POS sets differ from each other according to how much grammatical information they have: the first set has maximal syntactic and morphological information which possibly affects prosodic phrasing, and the third set has minimal one. We hand-labelled 150 sentences using each of three POS sets and conducted perception test. Based on the results of the test, stochastic language modeling method was used to predict prosodic boundary strength. The results showed that the use of each POS set led to not too much different efficiency in the prediction, but the second set was a little more efficient than the other two. As far as the complexity in stochastic language modeling is concerned, however, the third set may be also preferable.

• 대한음성학회지:말소리
• /
• 제43호
• /
• pp.1-9
• /
• 2002

This study aims to model Korean prosodic phrasing using CART(classification and regression tree) method. Our data are limited to Korean read speech. We used 400 sentences made up of editorials, essays, novels and news scripts. Professional radio actress read 400sentences for about two hours. We used K-ToBI transcription system. For technical reason, original break indices 1,2 are merged into AP. Differ from original K-ToBI, we have three break index Zero, AP and IP. Linguistic information selected for this study is as follows: the number of syllables in ‘Eojeol’, the location of ‘Eojeol’ in sentence and part-of-speech(POS) of adjacent ‘Eojeol’s. We trained CART tree using above information as variables. Average accuracy of predicting NonIP(Zero and AP) and IP was 90.4% in training data and 88.5% in test data. Average prediction accuracy of Zero and AP was 79.7% in training data and 78.7% in test data.