• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2010.05a
• /
• pp.981-984
• /
• 2010

In this paper, we used Support Vector Machines(SVMs) as the learning method, one of Artificial Neural Network, to segregated from the continuous speech into phonemes, an initial, medial, and final sound, and then, performed continuous speech recognition from it. A Decision boundary of phoneme is determined by algorithm with maximum frequency in a short interval. Speech recognition process is performed by Continuous Hidden Markov Model(CHMM), and we compared it with another phoneme segregated from the eye-measurement. From the simulation results, we confirmed that the method, SVMs, we proposed is more effective in an initial sound than Gaussian Mixture Models(GMMs).

• The Journal of the Acoustical Society of Korea
• /
• v.17 no.6
• /
• 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.

• The Journal of the Korea Contents Association
• /
• v.12 no.2
• /
• pp.105-115
• /
• 2012

In this paper, we propose a correction method using phoneme unit segmentation to solve misrecognition of Korean Texts in signboard images using weighted Disassemble Levenshtein Distance. The proposed method calculates distances of recognized texts which are segmented into phoneme units and detects the best matched texts from signboard text database. For verifying the efficiency of the proposed method, a database dictionary is built using 1.3 million words of nationwide signboard through removing duplicated words. We compared the proposed method to Levenshtein Distance and Disassemble Levenshtein Distance which are common representative text string comparison algorithms. As a result, the proposed method based on weighted Disassemble Levenshtein Distance represents an improvement in recognition rates 29.85% and 6% on average compared to that of conventional methods, respectively.

• The Journal of the Acoustical Society of Korea
• /
• v.31 no.3
• /
• pp.170-178
• /
• 2012

In the paper, phonetic decision tree of the triphone unit was built for the phoneme-based speech recognition of 640 stations which run by the Korail. The clustering rate was determined by Pearson and Regression analysis to decide threshold used in node splitting. Using the determined the clustering rate, thresholds are automatically decided by the threshold value according to the average clustering rate. In the recognition experiments for verifying the proposed method, the performance improved 1.4~2.3 % absolutely than that of the baseline system.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2003.05b
• /
• pp.366-369
• /
• 2003

본 연구에서는 강건한 문맥의존 음향모델을 작성하기 위한 기초적인 연구로서 문맥환경과 상태수의 변화에 따른 음향모델의 성능을 고찰하고자 한다. 음성은 시간함수로 표현되며 음절, 단어, 연속음성을 발성할때 자음과 모음에 따라 발성시간에 차이가 있으며 음성인식의 최소 인식단위로 널리 사용되는 음소의 앞과 뒤에 오는 문맥환경에 따라 인식성능에 많은 차이를 보이고 있다. 따라서 본 연구에서는 시간의 변화(상태수의 변화)와 상태분할 과정에서 문맥환경의 변화를 고려하여 다양한 형태의 문맥의존 음향모델을 작성하였다. 모델학습은 음소결정트리 기반 SSS 알고리즘(Phonetic Decision Tree-based Successive State Splitting： PDT-555)을 이용하였다 PDT-SSS 알고리즘은 미지의 문맥정보를 해결하기 위해 문맥방향과 시간방향으로 목표 상태수에 도달할 때까지 상태분할을 수행하여 모델을 작성하는 방법이다. 본 연구에서 강건한 문맥의존 음향모델을 학습하기 위한 방법의 유효성을 확인하기 위해 국어공학센터의 452 단어를 대상으로 음소와 단어인식 실험을 수행하였다. 실험결과, 음성의 시간변이에 따른 모델의 상태수와 각 음소의 문맥환경에 따라 인식성능의 변화를 고찰할 수 있었다. 따라서 본 연구는 향후 음성인식 시스템의 강건한 문맥의존 음향모델을 작성하는데 유효할 것으로 기대된다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.17 no.5
• /
• pp.472-481
• /
• 1992

In this paper, we proposed a method of segmenting speech signal by neural network and its validity is proved by computer simulation. The neural network Is composed of multi layer perceptrons with one hidden layer. The matching accuracies of the proposed algorithm are measured for continuous vowel and place names. The resulting average matching accuracy is 100% for speaker-dependent case, 99.5% for speaker-independent case and 94.5% for each place name when the neural network 1,; trained for 6 place names simultaneously.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.1
• /
• pp.4-10
• /
• 2000

The work presented in this paper is about a postprocessor, which improves the performance of automatic speech segmentation system by correcting the phoneme boundary errors. We propose a postprocessor that reduces the range of errors in the auto labeled results that are ready to be used directly as synthesis unit. Starting from a baseline automatic segmentation system, our proposed postprocessor trains the features of hand labeled results using multi-layer perceptron(MLP) algorithm. Then, the auto labeled result combined with MLP postprocessor determines the new phoneme boundary. The details are as following. First, we select the feature sets of speech, based on the acoustic phonetic knowledge. And then we have adopted the MLP as pattern classifier because of its excellent nonlinear discrimination capability. Moreover, it is easy for MLP to reflect fully the various types of acoustic features appearing at the phoneme boundaries within a short time. At the last procedure, an appropriate feature set analyzed about each phonetic event is applied to our proposed postprocessor to compensate the phoneme boundary error. For phonetically rich sentences data, we have achieved 19.9 % improvement for the frame accuracy, comparing with the performance of plain automatic labeling system. Also, we could reduce the absolute error rate about 28.6%.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.18 no.3
• /
• pp.127-134
• /
• 2018

In this paper, we have developed the Auto Lip-Sync Maya plug-in for extracting Korean phonemes from voice data and text information based on Korean and produce high quality 3D lip-sync animation using divided phonemes. In the developed system, phoneme separation was classified into 8 vowels and 13 consonants used in Korean, referring to 49 phonemes provided by Microsoft Speech API engine SAPI. In addition, the pronunciation of vowels and consonants has variety Mouth Shapes, but the same Viseme can be applied to some identical ones. Based on this, we have developed Auto Lip-sync Maya Plug-in based on Python to enable lip-sync animation to be implemented automatically at once.

• The Journal of the Acoustical Society of Korea
• /
• v.25 no.7
• /
• pp.312-318
• /
• 2006

The conventional lip-synch system has a two-step process, speech segmentation and recognition. However, the difficulty of speech segmentation procedure and the inaccuracy of training data set due to the segmentation lead to a significant Performance degradation in the system. To cope with that, the connected vowel recognition method using Head-Body-Tail (HBT) model is proposed. The HBT model which is appropriate for handling relatively small sized vocabulary tasks reflects co-articulation effect efficiently. Moreover the 7 vowels are merged into 3 classes having similar lip shape while the system is optimized by employing a class dependent SCHMM structure. Additionally in both end sides of each word which has large variations, 8 components Gaussian mixture model is directly used to improve the ability of representation. Though the proposed method reveals similar performance with respect to the CHMM based on the HBT structure. the number of parameters is reduced by 33.92%. This reduction makes it a computationally efficient method enabling real time operation.

• Annual Conference on Human and Language Technology
• /
• 1995.10a
• /
• pp.30-34
• /
• 1995

신경망을 이용하는 연속 음성 인식에서 학습이라 함은 인위적으로 분할된 음성 데이타를 토대로 진행되는 것이 지배적이었다. 그러나 분할된 음성데이타를 마련하기 위해서는 많은 시간과 노력, 숙련 등을 요구할 뿐만아니라 그 자체가 인식도메인의 변화나 확장을 어렵게 하는 하나의 요인 되기도 한다. 그래서 분할된 음성데이타의 사용을 가급적 피하고 그러면서도 성능을 떨어뜨리지 않는 신경망 학습법들이 나타나고 있다. 본 논문에서는 학습된 인식기를 이용하여 자동으로 한국어 음성데이타를 분할한 후 그 분할된 데이타를 이용하여 다시 인식기를 재학습시켜나가는 반복 과정을 소개하고자 한다. 여기에는 TDNN이 인식기로 사용되며 인식단위는 음소이다. 학습은 cross-validation 기법을 이용하여 제어된다.