• ETRI Journal
• /
• 제33권6호
• /
• pp.914-923
• /
• 2011

This paper concerns robust and reliable speaker model training for text-independent speaker verification. The baseline speaker modeling approach is the Gaussian mixture model (GMM). In text-independent speaker verification, the amount of speech data may be different for speakers. However, we still wish the modeling approach to perform equally well for all speakers. Besides, the modeling technique must be least vulnerable against unseen data. A traditional approach for GMM training is expectation maximization (EM) method, which is known for its overfitting problem and its weakness in handling insufficient training data. To tackle these problems, variational approximation is proposed. Variational approaches are known to be robust against overtraining and data insufficiency. We evaluated the proposed approach on two different databases, namely KING and TFarsdat. The experiments show that the proposed approach improves the performance on TFarsdat and KING databases by 0.56% and 4.81%, respectively. Also, the experiments show that the variationally optimized GMM is more robust against noise and the verification error rate in noisy environments for TFarsdat dataset decreases by 1.52%.

• 한국음향학회지
• /
• 제20권7호
• /
• pp.69-75
• /
• 2001

화자확인에서 화자내 변이, 잡음환경, 그리고 학습환경과 인식 환경의 불일치는 화자확인 시스템이 실용화될 수 없는 가장 큰 원인이다. 본 연구에서는, 실제 환경에 강인한 화자 확인 시스템의 구현에 초점을 맞추어 음성 전처리 과정인 잡음환경에 강인한 끝점추출 알고리즘, 잡음제거 및 마이크특성 보상기법, LPG(Linear Predictive Coefficient)켑스트럼 가중치에 의한 화자간 변별력 향상 기법을 제안한다. 실험 결과, LPC잔차신호(residue)를 이용한 끝점추출 알고리즘을 사용한 경우 약 17.65% 가량의 끝점 추출 에러율을 향상시켰으며, 제안한 잡음제거 및 마이크특성 보상기법을 사용한 경우 다른 마이크 환경에서 화자 오인식율이 약 36.93% 가량 개선되었다. 또한, 제안한 LPC켑스트럼 가중치에 의한 화자간 변별력 향상 기법은 평균 화자 오인식율을 약 6.515% 향상시켰다.

• 대한음성학회지:말소리
• /
• 제63호
• /
• pp.153-170
• /
• 2007

It is well known that when there is an acoustic mismatch between the speech obtained during training and testing, the accuracy of speaker verification systems drastically deteriorates. This paper presents the use of MFCCs' histogram enhancement technique in order to improve the robustness of a speaker verification system. The technique transforms the features extracted from speech within an utterance such that their statistics conform to reference distributions. The reference distributions proposed in this paper are uniform distribution and beta distribution. The transformation modifies the contrast of MFCCs' histogram so that the performance of a speaker verification system is improved both in the clean training and testing environment and in the clean training and noisy testing environment.

• 한국음향학회지
• /
• 제28권3호
• /
• pp.268-272
• /
• 2009

본 논문에서는 화자확인을 위해서 가우시안혼합모델에 forgetting factor를 갖는 점진적 강인적응 방법을 제안하였다. 화자인식 시스템에서 적은 양의 데이터로 좋은 성능을 얻기 위하여 화자모델 적응방법이 사용되고 있다. 그러나, 현재 사용되고 있는 적응방법은 불규칙한 발성변화와 잡음 같은 이씨에 취약하고, 그것은 부정확한 화자모델을 만들 수 있다. 또한 시간이 지날수록 모델에 새로운 데이터가 적응되는 비율이 줄어들게 되는 문제점이 있다. 제안된 알고리즘은 가우시안혼합모델을 이용한 화자모델에서 이상치에 의한 왜곡과 새로운 데이터에 대한 적응 비율을 일정이상으로 유지할 수 있도록 하기 위하여 점진적 강인적응 방법을 제안하였다. 점진적 강인적응은 화자인식에서 적은 양의 데이터로 등록하고 테스트된 새로운 데이터로 모델을 적응시키는 방법이다. 실험결과는 7개월에 걸쳐서 수집된 데이터로부터 제안된 방법이 이상치에 강인하고 새로운 데이터의 적응 비율을 일정하게 유지시킴을 보였다.

• 말소리와 음성과학
• /
• 제2권3호
• /
• pp.127-134
• /
• 2010

Additive noise and channel mismatch strongly degrade the performance of speaker verification systems, as they distort the features of speech. In this paper a histogram transformation technique is presented to improve the robustness of text-independent speaker verification systems. The technique transforms the features extracted from speech such that their histogram is conformed to a reference distribution. The effect of different standard deviations for the reference distribution is investigated. Experimental results indicate that, in channel mismatched environments, the proposed technique offers significant improvements over existing techniques. We also verify performance improvement of the proposed method using statistics.

• 대한음성학회지:말소리
• /
• 제63호
• /
• pp.125-137
• /
• 2007

Recently, many techniques have been proposed to improve the noise robustness for speaker verification. In this paper, we consider the feature recombination technique in multi-band approach. In the conventional feature recombination for speaker verification, to compute the likelihoods of speaker models or universal background model, whole feature components are used. This computation method is not effective in a view point of multi-band approach. To deal with non-effectiveness of the conventional feature recombination technique, we introduce a subband likelihood computation, and propose a modified feature recombination using subband likelihoods. In decision step of speaker verification system in noise environments, a few very low likelihood scores of a speaker model or universal background model cause speaker verification system to make wrong decision. To overcome this problem, a reliable feature selection method is proposed. The low likelihood scores of unreliable feature are substituted by likelihood scores of the adaptive noise model. In here, this adaptive noise model is estimated by maximum a posteriori adaptation technique using noise features directly obtained from noisy test speech. The proposed method using subband-based reliable feature selection obtains better performance than conventional feature recombination system. The error reduction rate is more than 31 % compared with the feature recombination-based speaker verification system.

• 음성과학
• /
• 제14권2호
• /
• pp.127-135
• /
• 2007

Feature vectors of speech are generally extracted from whole frequency domain. The inherent character of a speaker is located in the low band or high band frequency. However, if the speech is corrupted by narrowband noise with concentrated energy, speaker verification performance is reduced as the individual characteristic is removed. In this paper, we propose a PCA Covariance Model based on the multiband to extract the robust feature vectors against the narrowband noise. First, we divide the overall frequency band into several subbands. Second, the correlation of feature vectors extracted independently from each subband is removed by PCA. The distance obtained from each subband has different distribution. To normalize against the different distribution, we moved the value into the normalized distribution through the mapping function. Finally, the represented value applying the weighting function is used for speaker verification. In the experiments, the proposed method shows better performance of the speaker verification and reduces the computation.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2003년도 5월 학술대회지
• /
• pp.137-140
• /
• 2003

In this paper, we propose a robust endpoint detection algorithm for speaker verification. Proposed algorithm uses energy and cepstral distance parameters, and it replaces the detected endpoints with endpoints of voiced speech, when the estimated signal-to-noise ratio (SNR) is low. Experimental results show that proposed algorithm is superior to energy-based endpoint detection algorithm.

• 한국음향학회지
• /
• 제19권3호
• /
• pp.45-50
• /
• 2000

본 논문에서는 하위단어에 기반한 전화선 채널에서의 어구 종속 화자 확인 시스템을 위한 음성 분할 알고리즘인, 파라미트릭 필터링에 기반한 델타 에너지를 제안한다. 제안한 알고리즘은 특정 밴드의 주파수를 기준으로 대역폭을 변화시키며 필터링한 후 델타 에너지를 이용하는 방법으로 다른 알고리즘에 비해 주변환경에 강인한 것으로 나타났다. 이를 이용해 음성을 하위단어로 분할하고, 각 하위단어를 이용해 화자의 성문을 모델링하였다. 제안한 알고리즘의 성능 평가를 위해 EER(Equal Error Rate)를 사용한다. 그 결과 단일 모델의 EER이 약 6.1%, 하위 단어 모델의 EER이 약 4.0%로 본 논문에서 제안한 알고리즘을 사용했을 때 약 2%의 성능이 향상되었다.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2002년도 11월 학술대회지
• /
• pp.179-182
• /
• 2002

The development of noise robust speech processing systems is becoming increasingly important as speech technology is currently widely applied in real world applications. Recently, to resolve such a noise problem, adaptive noise canceller(ANC) is frequently used, which is based upon adaptive filters. The adaptive recursive filters perform better than adaptive non-recursive filters due to the added poles, but the stability may be severely threatened. But these problems of adaptive recursive filters was solved by ACHARF algorithm. This paper presents a method which combines speaker verification system with ANC(Adaptive Noise Canceller) using the ACHARF algorithm. In the front-end stage, ANC is adopted to suppress the additive noise imposed on the speech signal. The results show that the performance of speaker verification system becomes better than before.