• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.08a
• /
• pp.209-213
• /
• 1998

50개의 담화단독 문장과 연속발성 문장을 대상으로 무장의 첫 어절과 마지막 어절에서의 첫 음절과 마지막 음절의 운율특징을 조사하였다. 이를 체계적으로 살펴 보기 위하여 각 어절에서의 마지막 음절의 음향변수에 대한 첫 음절의 음향변수의 비율을 얻은 뒤 이를 대상으로 하여 평균값과 분포를 구하였다. 지속시간의 경우 두 스타일 간에 주목할 만한 큰 차이점은 없었으나 담화 연속 문장의 문두에서 화자의 조음시간 프로그래밍이 약간 조화롭지 못함을 알 수 있었다. Fo는 마지막 어절 부분의 비율값이 두 스타일간 통계적으로 유의한 차이를 보였으며 운율자질로 기능할 수 있는 가능성을 보였다. 에너지는 Fo와 유사한 분포경향을 보인다. 문미 어절의 마지막 음절이 첫 음절의 약 85% 정도의 힘으로 발성됨을 알 수 있고, 담화 연속 발화의 마지막 어절에서 단독 발화문보다 상대적으로 강하게 조음되었음을 알 수 있었다.

• Phonetics and Speech Sciences
• /
• v.6 no.3
• /
• pp.57-62
• /
• 2014

The present study was performed to investigate and discuss the aerodynamic characteristics of whispered and normal speech during reading paragraph tasks. 39 normal females(18-23 yrs.) read 'Autumn' paragraph with whispered and normal phonation. Their readings were recorded and analyzed by 'Running Speech' in Phonatory Aerodynamic System(PAS) instrument. As results, during whispered speech, the total duration was longer and the numbers of inspiration were more frequently shown than normal speech. The Peak expiratory and inspiratory rate were higher in normal speech, but the expiratory and inspiratory volume were higher in whispered speech. By correlation analysis, both whispered and normal speech showed significantly high correlation between total duration and expiratory/inspiratory airflow duration; numbers of inspiration and inspiratory airflow duration; expiratory and inspiratory volume. These results show that whispered speech needs more respiratory effort but shows poorer aerodynamic efficacy during phonation than normal speech.

• Phonetics and Speech Sciences
• /
• v.11 no.3
• /
• pp.39-47
• /
• 2019

We propose a method to improve the performance of spontaneous speech recognizers by extending their phone set using speech data. In the proposed method, we first extract variable-length phoneme-level segments from broadcast speech signals, and convert them to fixed-length latent vectors using an long short-term memory (LSTM) classifier. We then cluster acoustically similar latent vectors and build a new phone set by choosing the number of clusters with the lowest Davies-Bouldin index. We also update the lexicon of the speech recognizer by choosing the pronunciation sequence of each word with the highest conditional probability. In order to analyze the acoustic characteristics of the new phone set, we visualize its spectral patterns and segment duration. Through speech recognition experiments using a larger training data set than our own previous work, we confirm that the new phone set yields better performance than the conventional phoneme-based and grapheme-based units in both spontaneous speech recognition and read speech recognition.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.3
• /
• pp.330-338
• /
• 2002

Spontaneous speech is ungrammatical as well as serious phonological variations, which make recognition extremely difficult, compared with read speech. In this paper, for conversational speech recognition, we analyze the transcriptions of the real conversational speech, and then classify the characteristics of conversational speech in the speech recognition aspect. Reflecting these features, we obtain the baseline system for conversational speech recognition. The classification consists of long duration of silence, disfluencies and phonological variations; each of them is classified with similar features. To deal with these characteristics, first, we update silence model and append a filled pause model, a garbage model; second, we append multiple phonetic transcriptions to lexicon for most frequent phonological variations. In our experiments, our baseline morpheme error rate (WER) is 31.65%; we obtain MER reductions such as 2.08% for silence and garbage model, 0.73% for filled pause model, and 0.73% for phonological variations. Finally, we obtain 27.92% MER for conversational speech recognition, which will be used as a baseline for further study.

• The Journal of the Acoustical Society of Korea
• /
• v.14 no.1
• /
• pp.56-64
• /
• 1995

In this paper we extracted control rules of synthetical pauses and syllable durations depending on pronunciation speed in Korean speech from the spoken sentences recorded by 18 professional announcers. Pause rules were divided into three categories : pause between sentences(PBS), pause between clauses(PBC) and pause between intonational phrases(PBI). From the analysis results it is found that comparing the slowly spoken sentence with the fast spoken sentence the duration difference between them is due to the synthetical pause increments, expecially, of PBS and PBC. In addition, it is also found that the increment ratio of the mean syllable duration Is low. On the other hand, PBI was not pronounced in the fast spoken sentences. PBI was pronounced at the pronunciation speed(PS) over some PS.

• MALSORI
• /
• no.35_36
• /
• pp.1-12
• /
• 1998

This study aims to investigate the prosodic characteristics of Korean discourse sentences, especially focusing the initial and final part of a sentence. 50 disourse sentences were read in two different styles; one, sentence by sentence, the other, continuous of all 50's. First, we tried to get two kinds of ratios from the acoustic results: first, ratio of the final syllable to the initial syllable in first word in a sentence; second, ratio of the final syllable to the initial syllable in last word in a sentence. We, then, calculated statistical values of the ratios including mean, standard deviation, minimum, maximum, and p-values in t-test. With respect to duration, there were little difference between two different styles. If any, we could see tiny unharmonious durational aspect in the initial of continuous reading. More concisely, there could be observed some deviation from standard. In case of F0, there was prominent statistical difference between ratios of last words in two styles. This difference might play a role as a prosodic feature. Energy seems to show similar pattern with that of F0. The results showed that final syllable in last word was pronounced with about 85 % of initial syllable in the same context and the last words in continuous speech were strongly articulated compared with those of sentence by sentence reading.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.06c
• /
• pp.99-102
• /
• 1998

본 연구에서는 연속음 인식과 합성을 위한 경계강도 예측 모델을 제안한다. 운율 경계 강도는 음성 합성에서는 운율구 사이의 휴지기의 길이 조절로 합성음의 자연도에 기여를 하고 연속음 인식에서는 인식과정에서 나타나는 후보문장의 선별 과정에 특징변수가 되어 인식률 향상에 큰 역할을 한다. 음성학적으로 발화된 문장은 큰 경계 단위로 볼 때 운율구 형태로 이루어졌다고 볼 수 있으며 구의 경계는 문장의 문법적인 특징과 관련을 지을 수 있게 된다. 본 논문에서는 운율 경계 강도 수준을 4로 하고 문법적인 특징으로는 트리구조 방법으로 결정된 오른쪽 가지의 수식의 깊이(rd)와 link grammar방법으로 결정된 음절수(syl), 연결거리(torig)를 bigram 모형과 결합하여 운율적 경계 강도를 예측한다. 예측 모형으로는 다중 회귀 모형과 Marcov 모형을 제안한다. 이들 모형으로 낭독체 200 문장에 대해 실험한 결과 76%로 경계 강도를 예측할 수 있었다.

• Proceedings of the KSPS conference
• /
• 2006.11a
• /
• pp.117-119
• /
• 2006

This paper discusses the importance of silent pauses in the perception of prosodic boundaries in Korean speech. It is suggested that in speech in general, and in particular in spontaneous speech, silent pauses are neither necessary nor sufficient for the perception of prosodic boundaries. In read speech, however, there is a high correlation between the presence of a pause and the perception of a boundary. An experiment was carried out to determine whether removing the silent boundary from an extract of speech had a significant effect on the perception of boundaries in Korean read speech. Results suggest that while the presence of a silent boundary slightly reinforces the perception of a prosodic boundary, subjects are in general capable of perceiving the boundary without the silent pause.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.21-24
• /
• 2002

본 논문은 ETRI 음성정보연구센터에서 추진하고 있는 공통음성 DB 구축에 관하여 기술한다. 총 3 년(2001 11-2004. 10) 동안 음성인식, 음성합성, 화자인식 등 다양한 용도의 음성 DB 를 수집할 예정이며, 1년차인 2002 년에는 총 14 종의 음성 DB 를 수집할 계획이다. 공통 음성 DB 는 다양한 통신망(마이크, 헤드셋, VoIP, 유무선 전화망), 지역, 성별, 발성환경(사무실, 지하철, 도로 등)을 고려하여 설계하였으며, 발성대상은 숫자, 단어, 문장이고, 발성방법은 자유발화, 대화체, 낭독체 등 다양한 스타일의 음성 DB 로 구성되어 있다. 이에 본 논문에서는 총 14 종에 해당하는 공통음성 DB 의 구축내역과 구축방안 및 DB 구축 일정에 관해 기술하고자 한다.

• Phonetics and Speech Sciences
• /
• v.13 no.4
• /
• pp.35-46
• /
• 2021

In this study, we examined the interactive aspects between pitch reduction phenomena considered a universal language phenomenon and native language interference in the production of L2 intonation performed by Chinese learners of Korean. To investigate their interaction, we conducted an acoustic analysis using acoustic measures such as pitch span, pitch level, pitch dynamic quotient, skewness, and kurtosis. In addition, the correlation between text comprehension and pitch was examined. The analyzed material consisted of four Korean discourses containing five and seven sentences of varying difficulty. Seven Korean native speakers and thirty Chinese learners who differed in their Korean proficiency participated in the production test. The results, for differences by language, showed that Chinese had a more expanded pitch span, and a higher pitch level than Korean. The analysis between groups showed that at the beginner and intermediate levels, pitch reduction was prominent, i.e., their Korean was characterized by a compressed pitch span, low pitch level, and less sentence internal pitch variation. Contrariwise, the pitch use of advanced speakers was most similar to Korean native speakers. There was no significant correlation between text difficulty and pitch use. Through this study, we observed that pitch reduction was more pronounced than native language interference in the phonetic layer.