• Proceedings of the Korean Information Science Society Conference
• /
• 2004.10b
• /
• pp.700-702
• /
• 2004

A local-step optimization method is proposed to supplement the global-step optimization methods which adopt online update mode of internal weights and error energy as stop criterion in learning of multilayer perceptrons (MLPs). This optimization method is applied to the standard online error backpropagation(EBP) and the performance is evaluated for a speaker verification system.

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 2004.07a
• /
• pp.206-206
• /
• 2004

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.04b
• /
• pp.619-621
• /
• 2002

MLP(multilayer perceptron)는 기존의 패턴인식 방법에 비해 몇 가지 이점을 제공하지만 학습에 비교적 많은 시간을 요구한다. 이 점은 화자증명 시스템의 인식방법으로서 MLP를 사용할 경우 등록시간이 길어지는 문제를 발생시킨다. 본 논문에서는 기존의 시스템에서 채택한 화자군집 방법을 응용하여 MLP 학습에 필요만 배경화자 수를 줄임으로써 화자등록 시간을 단축하는 방법을 제안한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2000.10b
• /
• pp.434-436
• /
• 2000

본 논문에서는 기존의 확률적 화자군집 모델을 MLP(multi-layer perceptron)로 구현하는 방법과 원형 화자군집 모델이 갖는 문제를 해결할 수정 모델을 제시한다. 화자군집 모델은 화자등록 시간에 민감한 실용 환경에서 중요한 의미를 지닌다. 본 연구에서 사용한 인식단위는 여러 음소계열에서 지속적인 부분을 추출한 지속음이므로 화자등록과 증명 단계에서 특정한 어구에 한정되지 않는 어구독립 방식을 채택한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.05a
• /
• pp.139-140
• /
• 2008

문장 독립 화자 확인에서 SVM을 위한 적응된 GMM을 바탕으로 특징을 추출함으로써 GMM과 SVM 사이의 새로운 접근 방식을 제안한다. 우수한 측정성으로 인해, 적응된 GMM은 SVM 화자 확인을 위한 대규모의 음성 데이터로부터 적은 양의, 전형적인 특징 벡터를 추출해오곤 했다. 이 새로운 접근방식을 사용함으로써, 제안된 화자 확인 시스템은 기존의 GMM-UBM 시스템보다 훨씬 나은 성능을 보였다.

• Journal of the Institute of Convergence Signal Processing
• /
• v.3 no.1
• /
• pp.8-14
• /
• 2002

In this paper, we realized a real-time text-independent speaker recognition system using gaussian mixture model, and applied frame level likelihood normalization method which shows its effects in verification system. The system has three parts as front-end, training, recognition. In front-end part, cepstral mean normalization and silence removal method were applied to consider speaker's speaking variations. In training, gaussian mixture model was used for speaker's acoustic feature modeling, and maximum likelihood estimation was used for GMM parameter optimization. In recognition, likelihood score was calculated with speaker models and test data at frame level. As test sentences, we used text-independent sentences. ETRI 445 and KLE 452 database were used for training and test, and cepstrum coefficient and regressive coefficient were used as feature parameters. The experiment results show that the frame-level likelihood method's recognition result is higher than conventional method's, independently the number of registered speakers.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.12 no.8
• /
• pp.1456-1462
• /
• 2008

In general, it is known a single biometric-based personal authentication has limitation to improve recognition rate due to weakness of individual recognition scheme. The recognition rate of face recognition system can be reduced by environmental factor such as illumination, while speaker verification system does not perform well with added surrounding noise. In this paper, a multi-modal biometric system composed of face and voice recognition system is proposed in order to improve the performance of the individual authentication system. The proposed empirical weight sum rule based on the reliability of the individual authentication system is applied to improve the performance of multi-modal biometrics. Since the proposed system is implemented using JAVA applet with security function, it can be utilized in the field of user authentication on the generic Web.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.6
• /
• pp.714-719
• /
• 2005

The paper attempts to design and implement the system which can remotely check visitors' speech or Image by a mobile phone. This system is designed to recognize who a visitor is through the automatic calling service, not through a short message, via the mobile phone, even when the home owner is outside. In general, door locks are controlled through the home Server, but it is more effective to control door locks by using DTMF signal from a real-time point of view. The technology suggested in this paper makes it possible to communicate between the visiter and the home owner by making a phone call to tile home owner's mobile phone automatically when the visiter visits the house even if the home owner is outside, and if necessary, it allows for the home owner to control the door lock remotely. Thanks to the system, the home owner is not restricted by time or space for checking the visitor's identification and controlling the door lock. In addition, the security system is improved by changing from the existing password form to the combination of password and speaker verification lot the verification procedure required for controlling the door lock and setting the environment under consideration of any disadvantages which may occur when the mobile Phone is lost. Also, any existing problems such as reconnection to tile network for controlling tile door lock are solved by controlling the door lock in real time by use of DTMF signal while on the phone.

• The Journal of the Acoustical Society of Korea
• /
• v.14 no.6
• /
• pp.13-20
• /
• 1995

In this paper, the thermoacoustic refrigerating system was implemented and its operation was experimentally verified. The system is composed of several parts ,4 inch midrange speaker, speaker housing, chamber, stack housing, stack of plates, heat exchangers, thin pipe and cavity. The system is filled with He gas at 10 bar and contains T-type thermocouples and condenser microphone for measuring the temperature and pressure inside, respectively. In addition, cooling water is used for protecting speaker from thermal destruction and cooling down the hot heat exchanger. For the experimental verification of the implemented refigerating system, electrical impedance and resonance characteristics were measured. The results showed that it was most efficient to drive the system at 340 Hz. When operated at 340 Hz, $30^\circ{C}$ environments and 50 electical watts, the temperature of the cold region decreased by $16^\circ{C}$. The dissatisfaction mainly comes from the incomplete thermal insulation of the cold region. We also pointed out some guidelines to improve the performance for later study.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.55-58
• /
• 2000

본 논문에서는 화자 확인 시스템의 등록과 확인 과정의 채널 환경 불일치로 성능이 저하되는 문제를 해결하기 위한 새로운 정규화 방법에 대해 설명한다. 제안된 방법은 첫째, 입력 음성으로부터 효과적으로 채널을 추정$\cdot$보상하고 둘째, 스코어 정규화 과정에서 사칭자 모델로서 사용되는 world모델과의 차이를 채널 추정 및 화자 모델 생성에 효과적으로 사용하는 것을 목표로 한다. 이를 위해 입력 음성의 켑스트럼과 HMM world 모델의 파라메터인 평균 켑스트럼과의 차이를 통해 음소열에 종속적인 채널 켑스트럼인 Phone-Dependent Difference Cepstrum을 추정한다. 한편 입력 음성의 음소열은 world모델의 스코어를 얻는 과정에서 함께 얻어질 수 있다. 채널 추정 실험 결과를 통해서 가장 일반적인 채널 정규화방법인 CMS에 의해 추정된 채널에 비해 실제 채널과 유사하며 화자 고유의 특성을 왜곡시키지 않는 채널 추정이 가능함을 확인할 수 있었다.