• Proceedings of the IEEK Conference
• /
• 2002.06d
• /
• pp.191-194
• /
• 2002

The final purpose of this paper is the enhancement of speech recognition rate under the matched telephone environment between training data and test data. To analyze the effect by the distortion of the changing telephone channel on every call, MFCC is used as the feature parameter and CMN, RTCN, and RASTA are used as channel compensation techniques. For each case, the variation of feature parameters of all phones is analyzed. And, we find recognition rates according to each compensation method using the continuous HMM recognizer, and examine the relationship between variation and recognition rate.

• Journal of the Korean Institute of Telematics and Electronics A
• /
• v.31A no.6
• /
• pp.159-168
• /
• 1994

Stochastic computation is adopted to reduce the silicon area of the multipliers in implementing neural network in VLSI. In addition to this advantage, the stochastic computation has inherent random errors which is required for implementing Boltzmann machine. This random noise is useful for the simulated annealing which is employed to achieve the global minimum for the Boltzmann Machine. In this paper, we propose a method to implement the Boltzmann machine with stochastic computation and discuss the addition problem in stochastic computation and its simulated annealing in detail. According to this analysis Boltzmann machine using stochastic computation is suitable for the pattern recognition/completion problems. We have verified these results through the simulations for XOR, full adder and digit recognition problems, which are typical of the pattern recognition/completion problems.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.455-457
• /
• 1999

필기 숫자열의 인식성능을 향상시키기 위해서는 물론 인식기 자체의 성능 개선도 필요하지만 인식기에서 필요로 하는 정보를 제공해주는 초기단계의 개선 역시 매우 중요하다. 낱자와는 달리 숫자열 인식에서는 인식기에서 필요한 단위로 입력 데이터를 분할해야만 하는데 잡영, 기울어짐, 접촉 등의 원인에 의해서 쉽게 분할해내기 어렵기 때문이다. 본 논문에서는 이런 문제점들을 극복하기 위한 방법들은 제시하였으며 NIST 숫자열 데이터에 적용해 본 결과 16%의 성능 향상을 보였다.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.06a
• /
• pp.707-712
• /
• 1998

This paper presents several softcomputing techniques such as neural networks, fuzzy logic and genetic algorithms : Neural networks as brain metaphor provide fundamental structure, fuzzy logic gives a possibility to utilize top-down knowledge from designer, and genetic algorithms as evolution metaphor determine several system parameters with the process of bottom up development. With these techniques, we develop a pattern recognizer which consists of multiple neural networks aggregated by fuzzy integral in which genetic algorithms determine the fuzzy density values. The experimental results with the problem of recognizing totally unconstrained handwritten numeral show that the performance of the proposed method is superior to that of conventional methods.

• Proceedings of the KIEE Conference
• /
• 1987.11a
• /
• pp.338-340
• /
• 1987

The speech recognition system usually consists of two modules, segmentation module and identification module. So, the performance of the system heavily depends on the segmentation accuracy and the segmentation unit. This paper is concerned with the agreeable features for segmentation in syllables. Total energy and two band width energy. (LE:4000-5000Hz and HE:900-3100Hz) are suitable cues for segmentation. And we testify it through the experiment using connected digit.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.04b
• /
• pp.610-612
• /
• 2002

본 논문은 오프라인 필기체 숫자 인식을 위하여 SOG* 세선화와 방향 코드 생성만으로 전처리를 단순화하여 효율을 높이는 새로운 방법을 제안한다. 본 실험의 객관적 검증을 위해 Concordia 대학교 등의 여러기관의 필기체 숫자 데이터베이스에 대하여 실험한 결과 98.85% 이상의 인식률을 나타내어 단순한 전처리로 높은 인식률을 얻음으로써 효율성이 높음을 알 수 있다.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1995.10b
• /
• pp.52-58
• /
• 1995

A shared-weight neural network that performed classification based on the features extracted with the fuzzy morphological operation is introduced. Learning rules for the structuring elements, degree of membership, and weighting factors are also precisely described. In application to handwritten digit recognition problem, the fuzzy morphological shared-weight neural network produced the results which are comparable to the state-of-art for this problem.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.44-47
• /
• 1999

본 논문에서는 한국어 연결숫자인식에 있어서 모델간의 변별력 향상을 통해 인식률을 높이기 위하여 음향학적 파라메터(Acousticparameter)를 사용하는 짓을 제안한다. 제안된 방법은 음성학적 지식에 근거하여 적절한 주파수 대역별 에너지의 비의 로그값을 추가적인 특징파라메터로 사용한다. 실험결과, 제안된 방법을 사용함으로써 기본 인식시스템에 비해 오류율이 최고 $46\%$ 정도 감소됨을 확인할 수 있었다. 그리고 채널보상 기술을 함께 적용함으로써 $69\%$ 정도의 오류율 감소를 얻었다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2002.05a
• /
• pp.171-174
• /
• 2002

음성 신호는 언어정보, 개인성, 감정 등의 여러 가지 정보를 포함한 음향학적인 신호인 동시에 가장 자연스럽고 널리 쓰이는 의사소통 수단의 하나이다. 본 연구에서는 저장된 음성 신호에서 추출한 특징 파라미터를 사용한 경우와 음성 특징파라미터에 입술 패턴에 대한 영상정보를 통시에 사용한 경우 DHMM과 신경망을 통하여 각각 인식률을 비교해 보았다. 그 결과 입술패턴에 대할 영상정보도 음성인식에 사용 할 수 있음을 알 수 있었다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.9C
• /
• pp.853-858
• /
• 2006

TMHMM(Tied Mixture Hidden Markov Model) is an important approach to reduce the number of free parameters in speech recognition. However, this model suffers from a degradation in recognition accuracy due to its GPDF (Gaussian Probability Density Function) clustering error. This paper proposes a clustering algorithm, called HCNN(Homogeneous Centroid Neural network), to cluster acoustic feature vectors in TMHMM. Moreover, the HCNN uses the heterogeneous distance measure to allocate more code vectors in the heterogeneous areas where probability densities of different states overlap each other. When applied to Korean digit isolated word recognition, the HCNN reduces the error rate by 9.39% over CNN clustering, and 14.63% over the traditional K-means clustering.