• MALSORI
• /
• no.46
• /
• pp.51-64
• /
• 2003

In order to produce high quality synthesized speech, it is very important to get an accurate grapheme-to-phoneme conversion and prosody model from texts using natural language processing. Robust preprocessing for non-Korean characters should also be required. In this paper, we analyzed Korean texts using a morphological analyzer, part-of-speech tagger and syntactic chunker. We present a new grapheme-to-phoneme conversion method for Korean using a hybrid method with a phonetic pattern dictionary and CCV (consonant vowel) LTS (letter to sound) rules, for unlimited vocabulary Korean TTS. We constructed a prosody model using a probabilistic method and decision tree-based method. The probabilistic method atone usually suffers from performance degradation due to inherent data sparseness problems. So we adopted tree-based error correction to overcome these training data limitations.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1984.12a
• /
• pp.57-61
• /
• 1984

In this study, the system is proposed which can be used for recognition of Koean single vowles "ㅏ, ㅓ, ㅗ, ㅜ, ㅡ, ㅣ, ㅐ, ㅔ, ㅚ,", and automatic recognition is processed using $\mu$-computer. 3 men of not-being-studied are participated in this experiment. Using the period of vowels, one part of the steady state is selected for high speed recognition, and amplitude comparison method, LPC, PARCOR, and Formant are used for parameter of recognition. Formant is obtained by peak picking method using LPC, and then vowels are recognized by amplitude comparison method, LPC, PARCOR, and Formant. As a result, Recognition rates are 90.1% for amplitude comparison method, 93.1% for LPC, 100% for PARCOR, 88.8% for using formant.

• Proceedings of the KIEE Conference
• /
• 1987.07b
• /
• pp.1107-1110
• /
• 1987

한국어 음소 인식 시스템의 구현을 위한 한 단체로써 한국어 모음의 분류를 위한 시스템을 설계하였다. 특징 파라미터로는 에너지, 영교차율 그리고 제 1, 제 2, 제 3 포르만트를 사용하였다. [아, 어, 오, 우, 으, 이, 에, 애]의 8개의 단모음에 해당하는 총 132개의 음성 데이터에 대하여 실험하였다. 각각의 모음에 대한 포르만트의 평균치를 구한 뒤 이들의 값을 표준 패턴으로 갖는 인식 시스템을 구현한다. 구현된 인식 시스템에 의하여 새로운 음성의 모음부가 어떤 모음인지를 인식하며 결정할 때의 거리 측정 방식으로는 선형분류 함수를 사용하였다. 132개의 모음에 대하여 62.9%의 인식율을 나타내었다.

• Proceedings of the KIEE Conference
• /
• 1999.11c
• /
• pp.727-729
• /
• 1999

We have developed a realtime phonetic typewriter implemented on IBM PC with sound card based on Windows 95. In this system, analyzing of speech signal, learning of neural network, labeling of output neurons and visualizing of recognition results are performed on realtime. The developing environment for speech processing is established by adding various functions, such as editing, saving, loading of speech data and 3-D or gray level displaying of spectrogram. Recognition experimental using Korean phone had a 71.42% for 13 basic consonant and 90.01% for 7 basic vowel accuracy.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.2E
• /
• pp.3-8
• /
• 2001

In this paper, we will examine a Korean large vocabulary continuous speech recognition (LVCSR) system for broadcast news speech. The combined vowel and implosive unit is included in a phone set together with other short phone units in order to obtain a longer unit acoustic model. The effect of this unit is compared with conventional phone units. The dictionary units for language processing are automatically extracted from eojeols appearing in transcriptions. Triphone models are used for acoustic modeling and a trigram model is used for language modeling. Among three major speaker groups in news broadcasts-anchors, journalists and people (those other than anchors or journalists, who are being interviewed), the speech of anchors and journalists, which has a lot of noise, was used for testing and recognition.

• Proceedings of the KIEE Conference
• /
• 1989.07a
• /
• pp.481-485
• /
• 1989

The object of this parer is on the design of an broad board classifier for connected. Korean digit. Many approaches have been applied in speech recognition systems: parametric vector quantization, dynamic programming and hiden Markov model. In the 80's the neural network method, which is expected to solve complex speech recognition problems, came bach. We have chosen the rule based system for our model. The phoneme-groups that we wish to classify are vowel_like, plosive_like fricative_like, and stop_like.The data used are 1380 connected digits spoken by three untrained male speakers. We have seen 91.5% classification rate.

• English Language & Literature Teaching
• /
• v.13 no.2
• /
• pp.1-23
• /
• 2007

The purpose of this research is to explore the effects of systematic teaching English pronunciation and listening in English. Focusing on phonemes and words in pairs and sentences, the sound systems of the English and Korean languages are dealt with in conjunction with the test data. This paper first discusses the systemic, or primary interference and the habitual, or secondary interference that hinder comprehension of certain English sounds. Second, the analysis of input and output test data on the contrasting vowels and consonants shows statistic significance in terms of the probability (p value) of t-test. Third, the comparative data by means of percentile of right answers on contrasting vowel and consonant sounds expound the different sound systems of the English and Korean languages. With this data, problems in pronunciation of and listening to English, and the factors that may cause these problems are analyzed so that they can be used as a guideline for a systematic approach in teaching English learners, thus leading to more satisfactory performance.

• Speech Sciences
• /
• v.13 no.3
• /
• pp.121-132
• /
• 2006

In Hangeul, the graphemes of syllables are organized in horizontal, vertical and mixed (both horizontal & vertical) orientations, and the graphemic position of consonant(s) and vowel(s) within a each syllable needs to be maintained within a square pattern. We investigated the characteristics of writing errors of 9 stroke patients with right hemisphere (RH) lesions and compared it to the performances of 15 normal subjects. The subjects were asked to write to dictation of 90 Korean syllables. One of the interesting findings was that our patients manifested visuospatial errors which are not commonly observed in other language-speaking (e.g., English) patients due to the unique syllabic organizations of Korean writing system. The prominent errors in the RH group could be explained by the impaired RH which normally controls the visuospatial functions.

• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.45-48
• /
• 2000

In the case of making large vocabulary speech recognition system, it is better to use the segment than the syllable or the word as the recognition unit. In this paper, we study on the basis of making mono-phone for Korean speech recognition. For experiments, we use the speech toolkit of OGI in U.S.A. The result shows that the recognition rate of :he case in which the diphthong is established as a single unit is superior to that of the case in which the diphthong is established as two units, i.e. a glide plus a vowel. And also, the recognition rate by the number of consonants is a little different.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.40 no.4
• /
• pp.360-373
• /
• 1991

Since 70's, efficient speech recognition methods such as HMM or DTW have been introduced primarily for speaker dependent isolated words. These methods however have confronted with difficulties in recognizing continuous speech. Since early 80's, there has been a growing awareness that neural networks might be more appropriate for English and Japanese phoneme recognition using neural networks. Dealing with only a part of vowel or consonant set, Korean phoneme recognition still remains on the elementary level. In this light, we develop a system based on neural networks which can recognize major Korean phonemes. Through experiments using two neural networks, SOFM and TDNN, we obtained remarkable results. Especially in the case of using TDNN, the recognition rate was estimated about 93.78% for training data and 89.83% for test data.