• 말소리와 음성과학
• /
• 제7권4호
• /
• pp.41-47
• /
• 2015

This paper examines whether the Global Vector model is applicable to Korean data as a universal learning algorithm. The main purpose of this study is to compare the global vector model (GloVe) with the word2vec models such as a continuous bag-of-words (CBOW) model and a skip-gram (SG) model. For this purpose, we conducted an experiment by employing an evaluation corpus consisting of 70 target words and 819 pairs of Korean words for word similarities and analogies, respectively. Results of the word similarity task indicated that the Pearson correlation coefficients of 0.3133 as compared with the human judgement in GloVe, 0.2637 in CBOW and 0.2177 in SG. The word analogy task showed that the overall accuracy rate of 67% in semantic and syntactic relations was obtained in GloVe, 66% in CBOW and 57% in SG.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2005년도 춘계 학술대회 발표논문집
• /
• pp.3-7
• /
• 2005

This research is for finding prototypes and characteristics of intonation found in ${\ulcorner}$-a/e, $t{\int}ijo$<${\lrcorner}$ and ${\ulcorner}$p/simnida${\lrcorner}$ among modern Korean predicate statements by constructing spoken corpus based on the current radio broadcast. So the result of the study is as follows. : (1) The construction of the balanced spoken corpus and the standard for boundary determination of rhythm are needed for the intonation model of speech synthesis. (2) Korean intonation units have the splited word tone which includes the nuclear tone and the pre-nuclear tone makes unclear tone more detailed. (3) I made man and woman intonation models individually through t-test of SPSS. (4) The standard intonation model is devided '-ajo'type and '-nida'type

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2007년도 한국음성과학회 공동학술대회 발표논문집
• /
• pp.311-314
• /
• 2007

Spoken dialog system development includes many laborious and inefficient tasks. Since there are many components such as speech recognizer, language understanding, dialog management and knowledge management in a spoken dialog system, a developer should take an effort to edit corpus and train each model separately. To reduce a cost for editting corpus and training each models, we need more systematic and efficent working environment. For the working environment, we propose DialogStudio as an spoken dialog system workbench.

• 말소리와 음성과학
• /
• 제9권2호
• /
• pp.9-22
• /
• 2017

The purpose of this paper is to report the whole process of developing Korean Standard Speech Database (KSS DB). This project is supported by SPO (Supreme Prosecutors' Office) research grant for three years from 2014 to 2016. KSS DB is designed to provide speech data for acoustic-phonetic and phonological studies and speaker recognition system. For the samples to represent the spoken Korean, sociolinguistic factors, such as region (9 regional dialects), age (5 age groups over 20) and gender (male and female) were considered. The goal of the project is to collect over 3,000 male and female speakers of nine regional dialects and five age groups employing direct and indirect methods. Speech samples of 3,191 speakers (2,829 speakers and 362 speakers using direct and indirect methods, respectively) are collected and databased. KSS DB designs to collect read and spontaneous speech samples from each speaker carrying out 5 speech tasks: three (pseudo-)spontaneous speech tasks (producing prolonged simple vowels, 28 blanked sentences and spontaneous talk) and two read speech tasks (reading 55 phonetically and phonologically rich sentences and reading three short passages). KSS DB includes a 16-bit, 44.1kHz speech waveform file and a orthographic file for each speech task.

• 한국컴퓨터정보학회논문지
• /
• 제15권7호
• /
• pp.91-98
• /
• 2010

대용량 음성 DB를 사용하는 음편접합 TTS는 부가적인 신호처리 기술을 거의 사용하지 않고, 문맥을 반영하는 여러 합성유닛들을 결합해 합성음을 생성하기 때문에 높은 자연성을 가진다는 장점이 있다. 그러나 자연성, 개인성, 어조, 감정구현 등에서 활용성을 높이기 위해서는 음성DB의 크기가 비례적으로 증가하게 되므로 음운환경과 음향적 특성이 유사한 다수의 음편들을 제거하여 음성DB의 크기를 감축하기 위한 연구가 필수적이다 본 논문에서는DB감축을 위해 결정 트리 기반의 새로운 음소 군집화 방법을 이용하여 한국어 TTS용 합성단위음편 데이터베이스 구축 방법을 제안한다. 그리고 클러스터링방법에 대한 성능 평가를 위해서 언어 처리기, 운율 처리기, 음편 선택기, 합성음 생성기, 합성단위 음편데이터베이스, 음성신호 출력기로 구성되는 한국어 TTS 기본 시스템을 이용하여 합성음을 생성하였고 트리 클러스터링 방법 CM1, CM2와 전체 DB (Full DB)와 감축된 DB(Reduced DB)의 4가지 조합별로 제작된 음편 데이터베이스를 이용하여 각 조합에 대한 MOS 테스트를 수행하였다. 실험결과 제안된 방법을 사용할 경우 전체 음성DB의 크기를 23%로 줄일 수 있었고, 청취실험 결과 높은 MOS를 보이므로 향후 소용량 DB TTS에 적용 가능성을 보였다.

• 말소리와 음성과학
• /
• 제1권4호
• /
• pp.85-94
• /
• 2009

This paper investigates the intonation patterns of Korean spontaneous speech through an analysis of four dialogues in the domain of travel planning. The speech corpus, which is a subset of spontaneous speech database recorded and distributed by ETRI, is labeled in APs and IPs based on K-ToBI system using Momel, an intonation stylization algorithm. It was found that unlike in English, a significant number of APs and IPs include hesitation lengthening, which is known to be a disfluency phenomenon due to speech planning. This paper also claims that the hesitation lengthening is different from the IP-final lengthening and that it should be categorized as a new category, as it greatly affects the intonation patterns of the language. Except for the fact that 19.09% of APs show hesitation lengthening, the spontaneous speech shows the same AP patterns as in read speech with higher frequency of falling patterns such as LHL in comparison with read speech which show more LH and LHLH patterns. The IP boundary tones of spontaneous speech, showing the same five patterns such as L%, HL%, LHL%, H%, LH% as in read speech, show higher frequency of rising patterns (H% and LH%) and contour tones (HL%, LH%, LHL%) while read speech on the contrary shows higher frequency of falling patterns and simple tones at the end of IPs.

• 한국음향학회지
• /
• 제18권8호
• /
• pp.75-83
• /
• 1999

본 논문에서는 음성합성의 억양 제어를 위한 새로운 F0 궤적 모델을 제안한다. 제안한 모델은 발성된 문장의 F0 궤적을 중첩가산되는 사건들로 분해하고, 각 사건들을 가우시안 종모양의 사건함수로 모델링한다. 그리고 제안한 모델을 위한 파라미터 추정 알고리즘을 제시한다. 제안한 모델은 특정한 음운론적 지식에 기반하지 않았으며, F0 궤적의 분석단계와 합성단계에 모두 사용 가능하다. 제안한 모델의 성능평가를 위해 다양한 장르에서 추출한 여러 형태의 500문장의 코퍼스를 구축하고, 이를 전문 아나운서에게 발성하게 하여 구축한 음성코퍼스로 실험한 결과, 원음성의 F0 궤적과 제안한 모델에 의해 합성된 F0 궤적의 평균 제곱 오류근이 7.87Hz이었다.

• ETRI Journal
• /
• 제36권5호
• /
• pp.714-720
• /
• 2014

In this paper, we propose the use of data-driven probabilistic utterance-level decision logic to improve Weighted Finite State Transducer (WFST)-based endpoint detection. In general, endpoint detection is dealt with using two cascaded decision processes. The first process is frame-level speech/non-speech classification based on statistical hypothesis testing, and the second process is a heuristic-knowledge-based utterance-level speech boundary decision. To handle these two processes within a unified framework, we propose a WFST-based approach. However, a WFST-based approach has the same limitations as conventional approaches in that the utterance-level decision is based on heuristic knowledge and the decision parameters are tuned sequentially. Therefore, to obtain decision knowledge from a speech corpus and optimize the parameters at the same time, we propose the use of data-driven probabilistic utterance-level decision logic. The proposed method reduces the average detection failure rate by about 14% for various noisy-speech corpora collected for an endpoint detection evaluation.

• 대한음성학회지:말소리
• /
• 제56호
• /
• pp.225-235
• /
• 2005

Speech unit concatenation with a large database is presently the most popular method for speech synthesis. In this approach, the mismatches at the unit boundaries are unavoidable and become one of the reasons for quality degradation. This paper proposes an algorithm to reduce undesired discontinuities between the subsequent units. Optimal matching points are calculated in two steps. Firstly, the fullback-Leibler distance measurement is utilized for the spectral matching, then the unit sliding and the overlap windowing are used for the waveform matching. The proposed algorithm is implemented for the corpus-based unit concatenating Korean text-to-speech system that has an automatically labeled database. Experimental results show that our algorithm is fairly better than the raw concatenation or the overlap smoothing method.

• The Journal of the Acoustical Society of Korea
• /
• 제20권4E호
• /
• pp.31-38
• /
• 2001

A large-scale synthesis database for a unit selection based synthesis method usually retains redundant synthesis unit instances, which are useless to the synthetic speech quality. In this paper, to eliminate those instances from the synthesis database, we proposed a new pruning method called weighted vector quantization (WVQ). The WVQ reflects relative importance of each synthesis unit instance when clustering the similar instances using vector quantization (VQ) technique. The proposed method was compared with two conventional pruning methods through the objective and subjective evaluations of the synthetic speech quality: one to simply limit maximum number of instance, and the other based on normal VQ-based clustering. The proposed method showed the best performance under 50% reduction rates. Over 50% of reduction rates, the synthetic speech quality is not seriously but perceptibly degraded. Using the proposed method, the synthesis database can be efficiently reduced without serious degradation of the synthetic speech quality.