Improved normalization without recourse to an impostor database for speaker verification

Hébert, Matthieu; Peters, S.D.

doi:10.1109/icassp.2000.859184

Cited by 3 publications

(3 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on these observations and inspired by the work described in [11], we explored the effect of slightly perturbing the mean of the target-dependent Gaussian using the target-independent Gaussian.…”

Section: Lrementioning

confidence: 99%

Gaussian Backend design for open-set language detection

BenZeghiba

Gauvain

Lamel

2009

2009 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

This paper proposes a new approach to the challenging open-set language detection task. Most state-of-the-art approaches make use of data sources with several out-of-set languages to model such languages. In the proposed approach, no additional data from out-ofset languages is required, only date from the target languages is used. Experiments are conducted using the LRE-05 and the LRE-07 evaluation data sets with the 30s condition. A Cavg of 4.5% and 3.4% is obtained on these data set, respectively. These results are comparable with other reported results.

show abstract

Section: Lrementioning

confidence: 99%

Gaussian Backend design for open-set language detection

BenZeghiba

Gauvain

Lamel

2009

2009 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

show abstract

“…However, this model might not be optimal. To improve the competitiveness of this model, a previous study [11] proposed the use of a modified normalizing model (MNM) determined by perturbing the inferred background model using the enrollment data to reflect the lexical content of the speaker's password. In another study [12], the authors proposed the use of the speaker enrollment data to (1) train a background model with fewer number of parameters compared to the speaker model or (2) perturbing the temporal information by reversing the state order of the previously trained background model.…”

Section: Speaker Verification 71 Score Normalizationmentioning

confidence: 99%

“…In single reference modeling approach, the two log likelihood ratios LLR s in (10) and LLR u in (11) are estimated as follows:…”

Section: Single Reference Approachmentioning

confidence: 99%

User-customized password speaker verification using multiple reference and background models

BenZeghiba

Bourlard

2006

Speech Communication

View full text Add to dashboard Cite

Abstract. This paper discusses and optimizes an HMM/GMM based User-Customized Password Speaker Verification (UCP-SV) system. Unlike text-dependent speaker verification, in UCP-SV systems, customers can choose their own passwords with no lexical constraints. The password has to be pronounced a few times during the enrollment step to create a customer dependent model. Although potentially more "user-friendly", such systems are less understood and actually exhibit several practical issues, including automatic HMM inference, speaker adaptation, and efficient likelihood normalization. In our case, HMM inference (HMM topology) is performed using hybrid HMM/MLP systems, while the parameters of the inferred model, as well as their adaptation, will use GMMs. However, the evaluation of a UCP-SV baseline system shows that the background model used for likelihood normalization is the main difficulty. Therefore, to circumvent this problem, the main contribution of the paper is to investigate the use of multiple reference models for customer acoustic modeling and multiple background models for likelihood normalization. In this framework, several scoring techniques are investigated, such as Dynamic Model Selection (DMS) and fusion techniques. Results on two different experimental protocols show that an appropriate selection criteria for customer and background models can improve significantly the UCP-SV performance, making the UCP-SV system quite competitive with a text-dependent SV system. Finally, as customers' passwords are short, a comparative experiment using the conventional GMM-UBM text-independent approach is also conducted. 2 IDIAP-RR 04-41

show abstract