• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 1999년도 가을 학술발표논문집 Vol.26 No.2 (2)
• /
• pp.539-541
• /
• 1999

기존의 문장종속형 화자인식 기법에서는 음성 신호의 각 분석 프레임이 같은 기여도를 갖는 것으로 간주한다. 화자인식 시스템의 성능향상을 위해서는 음운정보보다는 인식의 단서가 되는 화자의 개인성 정보가 잘 반영되도록 하는 것이 중요하다. 본 논문에서는 HMM (hidden Markov model)을 기반으로 한 문장종속형 화자확인 시스템의 성능향상을 위해 프레임별로 인식의 단서가 되는 개인성 정보의 양을 측정하는 방법과, 이를 화자확인 시스템에 적용하는 기법을 제안한다. 제안한 방법을 적용한 결과, 기존의 우도비(likelihood ratio) 정규화 점수를 사용하는 방법에 비해 동일오류율(EER; equal error rate)을 평균 34% 감소시켜 인식율 향상을 얻을 수 있다.

• 말소리와 음성과학
• /
• 제4권3호
• /
• pp.95-101
• /
• 2012

In this paper, we introduce the automatic speaker identification system 'SPO(Supreme Prosecutors Office) Verifier'. SPO Verifier is a GMM(Gaussian mixture model)-UBM(universal background model) based automatic speaker recognition system and has been developed using Korean speakers' utterances. This system uses a channel compensation algorithm to compensate recording device characteristics. The system can give the users the ability to manage reference models with utterances from various environments to get more accurate recognition results. To evaluate the performance of SPO Verifier on Korean speakers, we compared this system with one of the most widely used commercial systems in the forensic field. The results showed that SPO Verifier shows lower EER(equal error rate) than that of the commercial system.

• 말소리와 음성과학
• /
• 제5권1호
• /
• pp.11-16
• /
• 2013

This paper proposes robust feature extraction for accurate voice activity detection (VAD). VAD is one of the principal modules for speech signal processing such as speech codec, speech enhancement, and speech recognition. Noisy environments contain nonstationary noises causing the accuracy of the VAD to drastically decline because the fluctuation of features in the noise intervals results in increased false alarm rates. In this paper, in order to improve the VAD performance, harmonic-weighted energy is proposed. This feature extraction method focuses on voiced speech intervals and weighted harmonic-to-noise ratios to determine the amount of the harmonicity to frame energy. For performance evaluation, the receiver operating characteristic curves and equal error rate are measured.

• 한국정보통신학회논문지
• /
• 제7권4호
• /
• pp.646-651
• /
• 2003

본 논문에서는 IMT2000 및 위성 DVB등 여러 규격에서 표준으로 채택된 터보 부호에서 사용되는 터보 인터리버의 성능에 대하여 비교분석하였다. BER 성능평가를 위해 레일리 페이딩 채널에서 고정소수점방식으로 컴퓨터 모의실험을 수행하였다. 공정한 비교를 위해 가능한한 인터리버의 크기를 같게 설정하였다. 모의실험결과 인터리버의 크기가 클수록 W-CDMA에서 사용하는 인터리버가 CDMA2000 및 위성 DVB용으로 사용하는 터보 인터리버에 비해 성능이 우수하다는 것을 확인하였다. 이러한 현상은 인터리버의 크기가 증가할 수록 더욱 두드러졌다.

• Journal of Information Processing Systems
• /
• 제14권2호
• /
• pp.523-538
• /
• 2018

Keystroke dynamics user authentication is a behavior-based authentication method which analyzes patterns in how a user enters passwords and PINs to authenticate the user. Even if a password or PIN is revealed to another user, it analyzes the input pattern to authenticate the user; hence, it can compensate for the drawbacks of knowledge-based (what you know) authentication. However, users' input patterns are not always fixed, and each user's touch method is different. Therefore, there are limitations to extracting the same features for all users to create a user's pattern and perform authentication. In this study, we perform experiments to examine the changes in user authentication performance when using feature vectors customized for each user versus using all features. User customized features show a mean improvement of over 6% in error equal rate, as compared to when all features are used.

• Journal of information and communication convergence engineering
• /
• 제18권2호
• /
• pp.132-146
• /
• 2020

Biometric technologies have become widely available in many different fields. However, biometric technologies using existing physical features such as fingerprints, facial features, irises, and veins must consider forgery and alterations targeting them through fraudulent physical characteristics such as fake fingerprints. Thus, a trend toward next-generation biometric technologies using behavioral biometrics of a living person, such as bio-signals and walking characteristics, has emerged. Accordingly, in this study, we developed a bio-signal authentication algorithm using electrocardiogram (ECG) signals, which are the most uniquely identifiable form of bio-signal available. When using ECG signals with our system, the personal identification and authentication accuracy are approximately 90% during a state of rest. When using fingerprints alone, the equal error rate (EER) is 0.243%; however, when fusing the scores of both the ECG signal and fingerprints, the EER decreases to 0.113% on average. In addition, as a function of detecting a presentation attack on a mobile phone, a method for rejecting a transaction when a fake fingerprint is applied was successfully implemented.

• Journal of Communications and Networks
• /
• 제8권1호
• /
• pp.38-48
• /
• 2006

A synchronous multicarrier (MC) code-division multiple access (CDMA) system using inverse fast Fourier transform (IFFT) and fast Fourier transform (FFT) for the downlink mobile communication system operating in a frequency selective Rayleigh fading channel is analyzed. Both carrier frequency offset and timing offset are considered in the analysis. Bit error rate performance of the system with both equal gain combining and maximum ratio combining are obtained. The performance is compared to that of the conventional system using correlation receiver. It is shown that when subcarrier number is large, the system using IFFT/FFT has nearly the same performance as the conventional one, while when the sub carrier number is small, the system using IFFT/FFT will suffer slightly worse performance in the presence of carrier frequency offset.

• 말소리와 음성과학
• /
• 제7권4호
• /
• pp.35-40
• /
• 2015

We propose a speech waveform forgery detection algorithm based on the flatness of frequency distribution. We devise a new measure of flatness which emphasizes the local change of the frequency distribution. Our measure calculates the sum of the differences between the energies of neighboring frequency bands. We compare the proposed measure with conventional flatness measures using a set of a large amount of test sounds. We also compare- the proposed method with conventional detection algorithms based on spectral distances. The results show that the proposed method gives lower equal error rate for the test set compared to the conventional methods.

• 한국음향학회:학술대회논문집
• /
• 한국음향학회 1999년도 학술발표대회 논문집 제18권 1호
• /
• pp.28-31
• /
• 1999

기존의 문장종속형 화자인식 방법들은 대부분 음성인식에서 사용되는 방법을 그대로 적용하기 때문에, 화자의 개인성 정보보다 음운정보에 더 민감한 단점이 있다. 화자인식 시스템의 성능향상을 위해서는 음운정보보다는 화자의 개인성 정보가 잘 반영되도록 하는 것이 중요하다. 본 논문에서는 HMM(hidden Maxkov model)을 기반으로 한 문장종속형 화자확인 시스템의 성능향상을 위한 관측확률 가중 반법을 제안한다. 먼저 주어진 학습자료에서 화자의 개인성이 잘 반영된 프레임들을 예측한다. 임의의 입력음성에 대한 인식점수는 화자의 특징이 잘 반영된 프레임의 관측확률에 가중치를 주어 구한다. 제안한 방법을 적용한 결과 기존의 우도비(likelihood ratio) 정규화 점수를 사용하는 방법에 비해 동일오류율(EER, equal error rate)을 $2\~3\%$정도 줄여 인식율 향상을 얻을 수 있었다.

• 센서학회지
• /
• 제20권6호
• /
• pp.388-392
• /
• 2011

In this paper, a face detection method for interactive TV control system using a new feature, edge histogram feature, with a support vector machine(SVM) in the near-infrared(NIR) images is proposed. The edge histogram feature is extracted using 16-directional edge intensity and a histogram. Compared to the previous method using local binary pattern(LBP) feature, the proposed method using edge histogram feature has better performance in both smaller feature size and lower equal error rate(EER) for face detection experiments in NIR databases.