• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.08a
• /
• pp.315-319
• /
• 1998

대어휘 연속음성인식을 위한 확률 발음사전 모델에 대해서 제안하였다. 제안된 확률 발음 사전은 연속음성과 같은 자연스런 발성에서 자주 발생되는 단어의 변이를 확률적인 subword-state로 이루어진 HMM으로 모델화 함으로써 단어의 발음 변이를 효과적으로 표현할 수 있으며, 단위 인식 시스템의 성능을 보다 높일 수 있도록 구성되었다. 확률 발음사전의 생성은 음성 자료와 음소 모델을 이용하여 단어 단위의 분할과 학습을 통해서 자동으로 생성되게 됨 음소와 같은 언어학적인 단위뿐만 아니라 PLU 이나 비언어학적인 인식 모델을 이용한 연속음성인식기에도 적용이 가능하다.연속음성인식실험결과 확률 발음사전을 사용함으로써 표준 발음 표기를 사용하는 인식 시스템에 비해 단어 오류율은 39.8%, 문장 오류율은 24.4%의 큰 폭으로 오류율을 감소시킬 수 있었다.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.6
• /
• pp.84-88
• /
• 1999

The accurate estimation of segmental duration is crucial for natural-sounding text-to-speech synthesis. For predicting Korean segmental durations, conventional methods utilized phonemic context, part-of-speech context and locational information in prosodic phrase. In this paper, the tonal information of phonemes is employed for more accurate prediction. After defining two non-boundary tones and six boundary tones, we annotated the tonal label on each syllable of 400 sentences. To predict segmental duration using tonal information, we constructed neural networks with a real-valued output node predicting phonemic duration and trained them by backpropagation algorithm. Experimental results showed that the proposed features are effective for predicting Korean segmental durations, and we got 0.863 correlation coefficient of the observed durations and predicted ones.

• Language Facts and Perspectives
• /
• v.44
• /
• pp.357-397
• /
• 2018

The purpose of this paper is to investigate the phonemic errors in Korean learner's spoken corpus. Through this, we tried to investigate the common errors and the errors in certain languages. The results of the analysis were as follows. First, Errors that distinguish three phonemes(plain sound, tense sound, aspiration sound) were high in all languages. In the middle phonemes, the most common errors in pronouncing 'ㅓ' in all languages. Second, the errors of each language are different. Comparing the ratios by position, Chinese characters had the most common errors with 50% in final phoneme, and the Japanese language showed equal errors in initial, middle, and end. In English, initial phoneme errors accounted for 58%. Vietnamese Learners showed intensive errors in the initial and final phoneme. Third, in addition to the phoneme errors, we also examined the allophone errors and foreign language pronunciation errors. The allophone errors are mainly concentrated in 'ㄹ', ​​and the pronunciation of the foreign language is mainly used in the source language or the native language of the learners. This paper analyzes the phoneme errors in the Learner's spoken language through the spoken corpus data with representative and annotation consistency. Through this study, we could compare the difference of phoneme errors of Main Korean learners.

• Journal of Digital Convergence
• /
• v.14 no.6
• /
• pp.499-511
• /
• 2016

The aim of this study was to find out effect of the orthographic representation on speech sound segmentation performance. Children's performances of the orthographic representation task and the speech sound segmentation task had positive correlation in words of phoneme-grapheme correspondence and negative correlation in words of phoneme-grapheme non-correspondence. In the case of words of phoneme-grapheme correspondence, there was no difference in performance ability between orthographic representation high level group and low level group, while in the case of words of phoneme-grapheme non-correspondence, the low level group's performance was significantly better than the high level group's. The most frequent errors of both groups were orthographic conversion errors and such errors were significantly more noticeable in the high level group. This study suggests that from the time of learning orthographic knowledge, children utilize orthographic knowledge for the performance of phonological awareness tasks.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.08a
• /
• pp.413-418
• /
• 1998

음성 합성 시스템의 한 모듈로서 한국어 문자열을 음소열로 자동 변환하는 시스템을 구현하였다. 문자열을 음소열로 변환할 때에는 한국어 음운현상에 대한 체계적인 분석 과정이 필요하다. 한국어의 음운 변화 현상은 단일 형태소 내부와 여러 형태소가 결합하여 한 어절을 이루는 경우 그 형태소 경계, 그리고 어절 경계에서 서로 다른 음운규칙이 적용된다. 따라서 언절이나 문장 등의 입력을 음소열로 변환하기 위해서는 형태소 분석, 태깅작업이 반드시 수행되어야 올바른 발음열을 유도할 수 있다. 본 논문에서 제안한 시스템은 한국어의 형태음운현상을 반영하기 위해 형태소 분석을 선행한 후, 한국어에서 빈번하게 발생하는 음운 변화 현상의 분석을 통해 정의된 음소 변동 규칙과 변이음 규칙을 선택적으로 적용하여 형태소, 어절, 언절 또는 문장 등의 다양한 형태의 입력에 대해 발음열을 생성한다. 기존의 연구에서 분리되어 있던 형태소 태거와 변환시스템을 통합하여 사용자 편의성을 높였으며 텍스트 기반의 형태소 분석기를 사용하기 때문에 원형이 복원되는 형태소들에 대한 처리 루틴을 두어 오류를 감소 시켰다.

• Journal of Korea Multimedia Society
• /
• v.14 no.1
• /
• pp.85-93
• /
• 2011

Previous existing vocabulary recognition programs calculate general vector values from a database, so they can not process phonemes that form during a search. And because they can not create a model for phoneme data, the accuracy of the Gaussian model can not secure. Therefore, in this paper, we recommend use of the Bhattacharyya distance measurement method based on the features of the phoneme-thus allowing us to improve the recognition rate by picking up accurate phonemes and minimizing recognition of similar and erroneous phonemes. We test the Gaussian model optimization through share continuous probability distribution, and we confirm the heighten recognition rate. The Bhattacharyya distance measurement method suggest in this paper reflect an average 1.9% improvement in performance compare to previous methods, and it has average 2.9% improvement based on reliability in recognition rate.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.5
• /
• pp.13-19
• /
• 2000

The blind segmentation method, which segments input speech data into recognition unit without any prior knowledge, plays an important role in continuous speech recognition system and corpus generation. As no prior knowledge is required, this method is rather simple to implement, but in general, it suffers from bad performance when compared to the knowledge-based segmentation method. In this paper, we introduce a method to improve the performance of a blind segmentation of Korean continuous speech by postprocessing the segment boundaries obtained from the blind segmentation. In the preprocessing stage, the candidate boundaries are extracted by a clustering technique based on the GLR(generalized likelihood ratio) distance measure. In the postprocessing stage, the final phoneme boundaries are selected from the candidates by utilizing a simple a priori knowledge on the syllabic structure of Korean, i.e., the maximum number of phonemes between any consecutive nuclei is limited. The experimental result was rather promising : the proposed method yields 25% reduction of insertion error rate compared that of the blind segmentation alone.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.10a
• /
• pp.532-534
• /
• 2003

코드로써 관리되어있지 않은 데이터베이스 내의 다양한 속성들이 시간이 흐름에 따라 정보로써 가치를 갖게 되면서. 비코드성 한글 데이터의 정형화에 대한 요구가 증가하고 있다. 정형화에 있어 한글의 특수성 중에 하나는 한글자료의 경우 KSC5601, CP949등을 사용하여 음절단위의 문자셋을 사용하여 음절단위로 저장 관리한다. 그런데 입력 시정에서는 자판기등을 이용하여 음소단위로 데이터를 입력하면서 발생하는 오류 및 비정형 데이터의 유입의 문제 등을 내포하고 있다. 이러한 문제를 해결하기 위하여 데이터의 저장단위인 음절이 아닌 음소 단위의 비교를 통하여 데이터를 정정하는 기법을 제안하고자 한다.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.2
• /
• pp.40-45
• /
• 2000

In speech corpus generation and speech recognition, it is sometimes needed to segment the input speech data without any prior knowledge. A method to accomplish this kind of segmentation, often called as blind segmentation, or acoustic segmentation, is to find boundaries which minimize the Euclidean distances among the feature vectors of each segments. However, the use of this metric alone is prone to errors because of the fluctuations or variations of the feature vectors within a segment. In this paper, we introduce the principal component analysis method to take the trend of feature vectors into consideration, so that the proposed distance measure be the distance between feature vectors and their projected points on the principal components. The proposed distance measure is applied in the LBDP(level building dynamic programming) algorithm for an experimentation of continuous speech segmentation. The result was rather promising, resulting in 3-6% reduction in deletion rate compared to the pure Euclidean measure.

• KIISE Transactions on Computing Practices
• /
• v.20 no.11
• /
• pp.579-585
• /
• 2014

A phoneme is the minimal unit of speech, and it plays a very important role in speech recognition. This paper proposes a novel method that can be used to recognize 'ㅅ', 'ㅈ', and 'ㅊ' among Korean phonemes. The proposed method is based on a volatility indicator and a turning point indicator that are calculated for each constituting block of the input speech signal. The volatility indicator is the sum of the differences between the values of each two samples adjacent in a block, and the turning point indicator is the number of extremal points at which the direction of the increment or decrement of the values of the sample are inverted in a block. A phoneme recognition algorithm combines the two indicators to finally determine the positions at which the three target phonemes mentioned above are recognized by utilizing optimized thresholds related with those indicators. The experimental results show that the proposed method can markedly reduce the error rate of the existing methods both in terms of the false reject rate and the false accept rate.