• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.3
• /
• pp.273-278
• /
• 2004

In this paper, the recurrent neural networks (RNN) is applied to compensate for HMM recognition algorithm, which is commonly used as main recognizer. Among these recurrent neural networks, the multi-layer recurrent neural prediction model (MRNPM), which allows operating in real-time, is used to implement learning and recognition, and HMM and MRNPM are used to design a hybrid-type main recognizer. After testing the designed speech recognition algorithm with Korean number pronunciations (13 words), which are hardly distinct, for its speech-independent recognition ratio, about 5% improvement was obtained comparing with existing HMM recognizers. Based on this result, only optimal (recognition) codes were extracted in the actual DSP (TMS320C6711) environment, and the embedded speech recognition system was implemented. Similarly, the implementation result of the embedded system showed more improved recognition system implementation than existing solid HMM recognition systems.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.2
• /
• pp.131-136
• /
• 2020

Wake Up Word (WUW) is a short utterance used to convert speech recognizer to recognition mode. The WUW defined by the user who actually use the speech recognizer is called user-defined WUW. In this paper, to recognize user-defined WUW, we construct traditional Gaussian Mixture Model-Hidden Markov Model (GMM-HMM), Linear Discriminant Analysis (LDA)-GMM-HMM and LDA-Deep Neural Network (DNN)-HMM based system and compare their performances. Also, to improve recognition accuracy of the WUW system, a threshold method is applied to each model, which significantly reduces the error rate of the WUW recognition and the rejection failure rate of non-WUW simultaneously. For LDA-DNN-HMM system, when the WUW error rate is 9.84 %, the rejection failure rate of non-WUW is 0.0058 %, which is about 4.82 times lower than the LDA-GMM-HMM system. These results demonstrate that LDA-DNN-HMM model developed in this paper proves to be highly effective for constructing user-defined WUW recognition system.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2007.11a
• /
• pp.443-446
• /
• 2007

음성신호는 잡음 또는 전송 채널의 특성에 의하여 왜곡되고, 왜곡된 음성은 음성인식 및 화자인식의 성능을 크게 저하시킨다. 이러한 문제점을 극복하기 위해 본 논문에서는 Gaussian mixture model (GMM)에 적용된 신호대잡음비 (SNR)기반 신뢰도 가중 기법[1][2]을 Hidden Markov model(HMM) 디코더에 변형하여 적용하였다. HMM 디코더 변형은 HMM 상태별 관측확률을 논문 [1]에서 제시된 신뢰도로 가중함으로써 이루어졌다. 제안한 방법의 성능을 확인하기 위해 ETRI에서 만든 한국어 화자인식용 휴대폰 음성 DB를 사용하여 문맥종속 화자식별 실험을 하였다. 실험결과 기존 방법에 비해 제안한 방법의 화자인식률이 크게 향상됨을 확인 할 수 있었다.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.23 no.9
• /
• pp.814-822
• /
• 2013

For the fault diagnosis of a mechanical system, pattern recognition methods have being used frequently in recent research. Hidden Markov model(HMM) and artificial neural network(ANN) are typical examples of pattern recognition methods employed for the fault diagnosis of a mechanical system. In this paper, a hybrid method that combines HMM and ANN for the fault diagnosis of a mechanical system is introduced. A rotating blade which is used for a wind turbine is employed for the fault diagnosis. Using the HMM/ANN hybrid model along with the numerical model of the rotating blade, the location and depth of a crack as well as its presence are identified. Also the effect of signal to noise ratio, crack location and crack size on the success rate of the identification is investigated.

• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.25-28
• /
• 1999

본 연구에서는 finite impulse response (FIR) 필터에 의해 인식기의 입력 특징벡터가 필터링되는 경우에 hidden Markov model (HMM) 파라미터를 적응시키는 새로운 기법을 제안한다. 제안한 적응 기법은 필터링에 의해 변환된 특징벡터에 대해 HMM 파라미터를 다시 학습시킬 필요가 없으며 주어진 FIR필터 계수만을 사용하여 HMM 파라미터를 적응시킬 수 있다. 개발된 FIR필터링에 대한 HMM 파라미터 적응 기법은 연속 숫자음 인식 실험에서 재학습 방법과 비교 실험한 결과 low-pass 필터의 경우에 재학습 방법과 비슷한 인식 성능을 나타내었다.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.11 no.3
• /
• pp.175-179
• /
• 2011

In this paper, we propose the MA(Model Adaption) HMM that to use speech enhancement and feature compensation. Normally voice reference data is not consider for real noise data. This method is not to use estimated noise but we use real life environment noise data. And we applied this contaminated data for recognition reference model that suitable for noise environment. MAHMM is combined with surround noise when generating reference patten. We improved voice recognition rate at mobile environment to use MAHMM.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.5 no.1
• /
• pp.1-6
• /
• 2012

In this paper, we propose a HMM(Hidden Markov Model)-based segmentation method to model shadows as well as foreground and background regions. The shadow of moving objects often keeps from visual tracking. We propose an HMM-based segmentation method which classifies each object in real time. In the case of traffic monitoring movies, the effectiveness of the proposed method was proved by experiments.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.34S no.4
• /
• pp.57-65
• /
• 1997

The purpose of this paper is to propose the HMM (hidden markov model) based on multi-observation sequence for the isolated word recognition. The proosed model generates the codebook of MSVQ by dividing each word into several sections followed by dividing training data into several sections. Then, we are to obtain the sequential value of multi-observation per each section by weighting the vectors of distance form lower values to higher ones. Thereafter, this the sequential with high probability value while in recognition. 146 DDD area names are selected as the vocabularies for the target recognition, and 10LPC cepstrum coefficients are used as the feature parameters. Besides the speech recognition experiments by way of the proposed model, for the comparison with it, the experiments by DP, MSVQ, and genral HMM are made with the same data under the same condition. The experiment results have shown that HMM based on multi-observation sequence proposed in this paper is proved superior to any other methods such as the ones using DP, MSVQ and general HMM models in recognition rate and time.

• Proceedings of the Korean Information Science Society Conference
• /
• 2008.06c
• /
• pp.472-477
• /
• 2008

본 논문에서는 은닉 마르코프 모델(HMM)을 기반한 대용량의 필기 한자 인식의 문제점인 시스템 리소스의 한계와 인식에 소요되는 많은 시간을 단축하기 위해 구조코드와 HMM에 최적화 된 클러스터링 알고리즘을 제안한다. 제안하는 클러스터링 알고리즘의 기본 개념은 훈련된 HMM를 대상으로 하고, HMM의 파라미터 수가 동일한 클래스에 대해서 클러스터를 구성하는 것이다. 또한 인식에 소요되는 시간을 줄이기 위해 2단계 클러스터모델 구조를 사용한다. 총 98,639 종류의 일본 한자를 대상으로 한 실험에서 평균 0.92 sec/char 인식 속도와 30순위 후보인식률 96.03%를 보임으로서 대용량 필기 한자 인식을 위한 좋은 방안이 될 것이라 기대한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.10b
• /
• pp.325-327
• /
• 2001

본 논문은 온라인 문자 인식을 연속 밀도 HMM의 구조의 최적화 문제를 다룬다. 최적이란 최소한의 모델 파라미터를 사용하여 최소한의 오류를 허용하는 것이라고 정의할 수 있다. 본 연구에서는 HMM 구조의 최적화를 위해 Bayesian 모델 선택 방법론을 사용한다. 먼저 잘 알려진 BIC(Bayesian Information Criterion)을 적용해보고, 그것을 HMM의 복잡한 구조에 적합하도록 본 논문에서 제안한 HBIC(HMM-Oriented BIC)와 비교해본다. BIC는 모델의 사전 확률 분포를 추정하지 않고 다변량 정규분포라고 가정하는데 비해 HBIC는 모델의 각 파라미터로부터 사전 확률을 추정한 후 그것들을 사용함으로써 더 좋은 결과를 얻도록 한다. 실험 결과 BIC와 HBIC 둘 다 기존 방법보다 모델의 파라미터 수를 현저히 감소시킴을 확인했고, HBIC가 BIC에 비해 더 적은 수의 파라미터를 사용해도 비슷한 인식률을 얻을 수 있었다.