Speaker recognition by location in the space of reference speakers

Mami, Yassine; Charlet, Delphine

doi:10.1016/j.specom.2005.06.014

Cited by 23 publications

(19 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Representing a speaker relative to other speakers is proposed in [154,218]. Each speaker model is presented as a combination of some reference models known as the anchor models.…”

Section: Other Modelsmentioning

confidence: 99%

An overview of text-independent speaker recognition: From features to supervectors

Kinnunen¹,

Li²

2010

Speech Communication

1,217

706

View full text Add to dashboard Cite

This paper gives an overview of automatic speaker recognition technology, with an emphasis on text-independent recognition. Speaker recognition has been studied actively for several decades. We give an overview of both the classical and the state-of-the-art methods. We start with the fundamentals of automatic speaker recognition, concerning feature extraction and speaker modeling. We elaborate advanced computational techniques to address robustness and session variability. The recent progress from vectors towards supervectors opens up a new area of exploration and represents a technology trend. We also provide an overview of this recent development and discuss the evaluation methodology of speaker recognition systems. We conclude the paper with discussion on future directions.

show abstract

“…Representing a speaker relative to other speakers is proposed in [154,218]. Each speaker model is presented as a combination of some reference models known as the anchor models.…”

Section: Other Modelsmentioning

confidence: 99%

An overview of text-independent speaker recognition: From features to supervectors

Kinnunen¹,

Li²

2010

Speech Communication

1,217

706

View full text Add to dashboard Cite

show abstract

“…This technique led to a major improvement in computational cost with only a small decrease in identification rate when used to reduce the size of the database prior to identification. More recent studies improved the method of anchor models by imposing constraints on the set of anchor models in order to derive an optimal projection space [29] and appropriate distance measures [30], [31]. Sakata et al [32] used a similar approach to create a secure fingerprint identification system, which did not store the images of the enrolled individuals.…”

Section: Indexing Using Match Scoresmentioning

confidence: 99%

Index Codes for Multibiometric Pattern Retrieval

Gyaourova

Ross

2012

IEEE Trans.Inform.Forensic Secur.

View full text Add to dashboard Cite

Abstract-In a biometric identification system, the identity corresponding to the input data (probe) is typically determined by comparing it against the templates of all identities in a database (gallery). Exhaustive matching against a large number of identities increases the response time of the system and may also reduce the accuracy of identification. One way to reduce the response time is by designing biometric templates that allow for rapid matching, as in the case of IrisCodes. An alternative approach is to limit the number of identities against which matching is performed based on criteria that are fast to evaluate. We propose a method for generating fixed-length codes for indexing biometric databases. An index code is constructed by computing match scores between a biometric image and a fixed set of reference images. Candidate identities are retrieved based on the similarity between the index code of the probe image and those of the identities in the database. The proposed technique can be easily extended to retrieve pertinent identities from multimodal databases. Experiments on a chimeric face and fingerprint bimodal database resulted in an 84% average reduction in the search space at a hit rate of 100%. These results suggest that the proposed indexing scheme has the potential to substantially reduce the response time without compromising the accuracy of identification.

show abstract

“…However, we found that direct fusion is often dominated by one particular LR, or it is limited by some inferior LRs. The concept of our methods is similar to that of the anchor modeling approach [20], [21] used in speaker indexing and speaker identification applications. The objective of the anchor modeling approach is to construct a speaker space based on a set of pre-trained representative models {A 1 ,A 2 ,…,A N }, called anchor models.…”

Section: A Direct Fusion Of Multiple Lrsmentioning

confidence: 99%

Using Kernel Discriminant Analysis to Improve the Characterization of the Alternative Hypothesis for Speaker Verification

Chao

Tsai

Wang

et al. 2008

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Abstract-Speaker verification can be viewed as a task of modeling and testing two hypotheses: the null hypothesis and the alternative hypothesis. Since the alternative hypothesis involves unknown impostors, it is usually hard to characterize a priori. In this paper, we propose improving the characterization of the alternative hypothesis by designing two decision functions based, respectively, on a weighted arithmetic combination and a weighted geometric combination of discriminative information derived from a set of pre-trained background models. The parameters associated with the combinations are then optimized using two kernel discriminant analysis techniques, namely, the Kernel Fisher Discriminant (KFD) and Support Vector Machine (SVM). The proposed approaches have two advantages over existing methods. The first is that they embed a trainable mechanism in the decision functions. The second is that they convert variable-length utterances into fixed-dimension characteristic vectors, which are easily processed by kernel discriminant analysis. The results of speaker-verification experiments conducted on two speech corpora show that the proposed methods outperform conventional likelihood ratio-based approaches.Index Terms-Kernel Fisher discriminant, likelihood ratio, speaker verification, support vector machine.

show abstract

Speaker recognition by location in the space of reference speakers

Cited by 23 publications

References 11 publications

An overview of text-independent speaker recognition: From features to supervectors

An overview of text-independent speaker recognition: From features to supervectors

Index Codes for Multibiometric Pattern Retrieval

Using Kernel Discriminant Analysis to Improve the Characterization of the Alternative Hypothesis for Speaker Verification

Contact Info

Product

Resources

About