channel compensation

In this paper we present a new method for text-independent speaker verification that combines segmental dynamic time warping (SDTW) and the d-vector approach. The d-vectors, generated from a feed forward deep neural network trained to distinguish between speakers, are used as features to perform alignment and hence calculate the overall distance between the enrolment and test utterances. We present results on the NIST 2008 data set for speaker verification where the proposed method outperforms the conventional i-vector baseline with PLDA scores and outperforms d-vector approach with local distances based on cosine and PLDA scores. Also score combination with the i-vector/PLDA baseline leads to significant gains over both methods.

remains an open problem that has not been satisfactorily solved by existing recognition techniques. In this paper, we tackle this problem using a variant of the recently proposed Probabilistic Linear Discriminant Analysis (PLDA). We show that simplified versions of the PLDA model, which are regularly used in the field of speaker recognition, rely on certain assumptions that not only result in a simpler PLDA model, but also reduce the computational load of the technique and -as indicated by our experimental assessments -improve recognition performance. Moreover, we show that, contrary to the general belief that PLDA-based methods produce well calibrated verification scores, score normalization techniques can still deliver significant performance gains, but only if nonparametric score normalization techniques are employed. Last but not least, we demonstrate the competitiveness of the simplified PLDA model for face recognition by comparing our results with the state-of-the-art results from the literature obtained on the second version of the large-scale Face Recognition Grand Challenge (FRGC) database.

Acoustic mismatch between the training and recognition conditions presents one of the serious challenges faced by speaker recognition researchers today. The goal of channel compensation is to achieve performance approaching that of a "matched condition" system while avoiding the need for a large amount of training data. It is important to ensure that the channel compensation algorithms in these systems compensate the channel variation instead of speaker variation. This paper addresses the problem of unsupervised compensation in which the features of a test utterance are transformed to fit the clean speaker model and gender-dependent background model. Specifically, a feature-based transformation is estimated based on the statistical difference between a test utterance and a composite acoustic model formed by combining the speaker and background models. By transforming the features to fit both models, the transformation is implicitly constrained. Experimental results based on the 2001 NIST evaluation set show that the proposed transformation approach achieves significant improvement in both equal error rate and minimum detection cost as compared to cepstral mean subtraction, Znorm and short-time Gaussianization.

Feature transformation aims to reduce the effects of channel-and handset-distortion in telephone-based speaker verification. This paper compares several feature transformation techniques and evaluates their verification performance and computation time under the 2000 NIST speaker recognition evaluation protocol. Techniques compared include feature mapping (FM), stochastic feature transformation (SFT), blind stochastic feature transformation (BSFT), feature warping (FW), and short-time Gaussianization (STG). The paper proposes a probabilistic feature mapping (PFM) in which the mapped features depend not only on the top-1 decoded Gaussian but also on the posterior probabilities of other Gaussians in the root model. The paper also proposes speeding up the computation of PFM and BSFT parameters by considering the top few Gaussians only. Results show that PFM performs slightly better than FM and that the fast approach can reduce computation time substantially. Among the approaches investigated, the fast BSFT (fBSFT) strikes a good balance between computational complexity and error rates, and FW and STG are the best in terms of error rates but with higher computational complexity. It was also found that fusion of the scores derived from systems using fBSFT and STG can reduce the error rate further. This study advocates that fBSFT, FW, and STG have the highest potential for robust speaker verification over telephone networks because they achieve good performance without any a priori knowledge of the communication channel.

Speaker Identification process is to identify a particular vocal cord from a set of existing speakers. In the speaker identification processes, unknown speaker voice sample targets each of the existing speakers present in the system and gives a predication. The predication may be more than one existing known speaker voice and is very close to the unknown speaker voice. The model is a Gaussian mixture model built by the extracted acoustic feature vectors from voice. The i-vector based dimension compression mapping function of the channel depended speaker, and super vector give better predicted scores according to cosine distance scoring associated with the order pair of speakers. In the order pair, the first coordinate is the unknown speaker i.e. test speaker, and the second coordinates is the existing known speaker i.e. target speaker. This paper presents the enhancement of the prediction based on i- vector in compare to the normalized set of predicted score. In the simulation, know...

Speaker verification in real-world applications sometimes deals with limited duration of enrollment and/or test data. MFCC-based i-vector systems have defined the state-of-the-art for speaker verification, but it is well known that they are less effective with short utterances. To address this issue, we propose a method to leverage the speaker specificity and stationarity of subglottal acoustics. First, we present a deep neural network (DNN) based approach to estimate subglottal features from speech signals. The approach involves training a DNN-regression model that maps the log filter-bank coefficients of a given speech signal to those of its corresponding subglottal signal. Cross-validation experiments on the WashU-UCLA corpus (which contains parallel recordings of speech and subglottal acoustics) show the effectiveness of our DNN-based estimation algorithm. The average correlation coefficient between the actual and estimated subglottal filter-bank coefficients is 0.9. A scorelevel fusion of MFCC and subglottal-feature systems in the ivector PLDA framework yields statistically-significant improvements over the MFCC-only baseline. On the NIST SRE 08 truncated 10sec-10sec and 5sec-5sec core evaluation tasks, the relative reduction in equal error rate ranges between 6 and 14% for the conditions tested with both microphone and telephone speech.

During late-2013 through mid-2014 NIST coordinated a special machine learning challenge based on the i-vector paradigm widely used by state-of-the-art speaker recognition systems. The i-vector challenge was run entirely online and used as source data fixed-length feature vectors projected into a low-dimensional space (i-vectors) rather than audio recordings. These changes made the challenge more readily accessible, enabled system comparison with consistency in the front-end and in the amount and type of training data, and facilitated exploration of many more approaches than would be possible in a single evaluation as traditionally run by NIST. Compared to the 2012 NIST Speaker Recognition Evaluation, the i-vector challenge saw approximately twice as many participants, and a nearly two orders of magnitude increase in the number of systems submitted for evaluation. Initial results indicate that the leading system achieved a relative improvement of approximately 38% over the baseline system.

Probabilistic linear discriminant analysis (PLDA) is a popular normalization approach for the i-vector model, and has delivered state-of-the-art performance in speaker recognition. A potential problem of the PLDA model, however, is that it essentially assumes Gaussian distributions over speaker vectors, which is not always true in practice. Additionally, the objective function is not directly related to the goal of the task, e.g., discriminating true speakers and imposters. In this paper, we propose a max-margin metric learning approach to solve the problems. It learns a linear transform with a criterion that the margin between target and imposter trials are maximized. Experiments conducted on the SRE08 core test show that compared to PLDA, the new approach can obtain comparable or even better performance, though the scoring is simply a cosine computation.

In speaker recognition, the mismatch between the enrollment and test utterances due to noise with different signal-to-noise ratios (SNRs) is a great challenge. Based on the observation that noise-level variability causes the i-vectors to form heterogeneous clusters, this paper proposes using an SNRaware deep neural network (DNN) to guide the training of PLDA mixture models. Specifically, given an i-vector, the SNR posterior probabilities produced by the DNN are used as the posteriors of indicator variables of the mixture model. As a result, the proposed model provides a more reasonable soft division of the i-vector space compared to the conventional mixture of PLDA. During verification, given a test trial, the marginal likelihoods from individual PLDA models are linearly combined by the posterior probabilities of SNR levels computed by the DNN. Experimental results for SNR mismatch tasks based on NIST 2012 SRE suggest that the proposed model is more effective than PLDA and conventional mixture of PLDA for handling heterogeneous corpora.

Language mismatch remains a major hindrance to the extensive deployment of speaker verification (SV) systems. Current language adaptation methods in SV mainly rely on linear projection in embedding space; i.e., adaptation is carried out after the speaker embeddings have been created, which underutilizes the powerful representation of deep neural networks. This paper proposes a maximum mean discrepancy (MMD) based framework for adapting deep neural network (DNN) speaker embedding across languages, featuring multilevel domain loss, separate batch normalization, and consistency regularization. We refer to the framework as MSC. We show that (1) minimizing domain discrepancy at both frame-and utterancelevels performs significantly better than at utterance-level alone; (2) separating the source-domain data from the target-domain in batch normalization improves adaptation performance; and (3) data augmentation can be utilized in the unlabelled target-domain through consistency regularization. By combining these findings, we achieve an EER of 8.69% and 7.95% in NIST SRE 2016 and 2018, respectively, which are significantly better than the previously proposed DNN adaptation methods. Our framework also works well with backend adaptation. By combining the proposed framework with backend adaptation, we achieve an 11.8% improvement over the backend adaptation in SRE18. When applying our framework to a 121-layer Densenet, we achieved an EER of 7.81% and 7.02% in NIST SRE 2016 and 2018, respectively.

Feature transformation aims to reduce the effects of channel-and handset-distortion in telephone-based speaker verification. This paper compares several feature transformation techniques and evaluates their verification performance and computation time under the 2000 NIST speaker recognition evaluation protocol. Techniques compared include feature mapping (FM), stochastic feature transformation (SFT), blind stochastic feature transformation (BSFT), feature warping (FW), and short-time Gaussianization (STG). The paper proposes a probabilistic feature mapping (PFM) in which the mapped features depend not only on the top-1 decoded Gaussian but also on the posterior probabilities of other Gaussians in the root model. The paper also proposes speeding up the computation of PFM and BSFT parameters by considering the top few Gaussians only. Results show that PFM performs slightly better than FM and that the fast approach can reduce computation time substantially. Among the approaches investigated, the fast BSFT (fBSFT) strikes a good balance between computational complexity and error rates, and FW and STG are the best in terms of error rates but with higher computational complexity. It was also found that fusion of the scores derived from systems using fBSFT and STG can reduce the error rate further. This study advocates that fBSFT, FW, and STG have the highest potential for robust speaker verification over telephone networks because they achieve good performance without any a priori knowledge of the communication channel.

This paper describes the speaker verification (SV) system submitted to the NIST 2016 speaker recognition evaluation (SRE) challenge by Indian Institute of Technology Guwahati (IITG) under the fixed training condition task. Various SV systems are developed following the idea-level collaboration with two other Indian institutions. Unlike the previous SREs, this time the focus was on developing SV system using non-target language speech data and a small amount unlabeled data from target language/dialects. For addressing these novel challenges, we tried exploring the fusion of systems created using different features, data conditioning, and classifiers. On NIST 2016 SRE evaluation data, the presented fused system resulted in actual detection cost function (actDCF) and equal error rate (EER) of 0.81 and 12.91%, respectively. Post-evaluation, we explored a recently proposed pairwise support vector machine classifier and applied adaptive S-norm to the decision scores before fusion. With these changes, the final system achieves the actDCF and EER of 0.67 and 11.63%, respectively.

Speaker de-identification approaches must accomplish three main goals: universality, naturalness and reversibility. The main drawback of the traditional approach to speaker de-identification using voice conversion techniques is its lack of universality, since a parallel corpus between the input and target speakers is necessary to train the conversion parameters. It is possible to make use of a synthetic target to overcome this issue, but this harms the naturalness of the resulting de-identified speech. Hence, a technique is proposed in this paper in which a pool of pre-trained transformations between a set of speakers is used as follows: given a new user to de-identify, its most similar speaker in this set of speakers is chosen as the source speaker, and the speaker that is the most dissimilar to the source speaker is chosen as the target speaker. Speaker similarity is measured using the i-vector paradigm, which is usually employed as an objective measure of speaker de-identification performance, leading to a system with high de-identification accuracy. The transformation method is based on frequency warping and amplitude scaling, in order to obtain natural sounding speech while masking the identity of the speaker. In addition, compared to other voice conversion approaches, the proposed method is easily reversible. Experiments were conducted on Albayzin database, and performance was evaluated in terms of objective and subjective measures. These results showed a high success when de-identifying speech, as well as a great naturalness of the transformed voices. In addition, when making the transformation parameters available to a trusted holder, it is possible to invert the de-identification procedure, hence recovering the original speaker identity. The computational cost of the proposed approach is small, making it possible to produce de-identified speech in real-time with a high level of naturalness.

I-vector extraction and Probabilistic Linear Discriminant Analysis (PLDA) has become the state-of-the-art configuration for speaker verification. Recently, Gaussian-PLDA has been improved by a preliminary length normalization of i-vectors. This normalization, known to increase the Gaussianity of the i-vector distribution, also improves performance of systems based on standard Linear Discriminant Analysis (LDA) and "two-covariance model" scoring. But this technique follows a standardization of the i-vectors (centering and whitening ivectors based on the first and second order moments of the development data). We propose in this paper two techniques of normalization based on total, between-and within-speaker variance spectra 1. These "spectral" techniques both normalize the i-vectors length for Gaussianity, but the first adapts the ivectors representation to a speaker recognition system based on LDA and two-covariance scoring when the second adapts it to a Gaussian-PLDA model. Significant performance improvements are demonstrated on the male and female telephone portion of NIST SRE 2010.

There are many factors affecting the variability of an i-vector extracted from a speech segment such as the acoustic content, segment duration, handset type and background noise. The language being spoken is one of the sources of variation which has received limited focus due to the lack of multilingual resources available. Consequently, the discrimination performance is much lower under multilingual trial condition. Standard session-compensation techniques such as Within-Class Covariance Normalization (WCCN), Linear Discriminant Analysis (LDA) and Probabilistic LDA (PLDA) cannot robustly compensate for language source of variation as the amount of data is limited to represent such variability. Source normalization technique which was developed to compensate for speech-source-variation, offered superior performance in cross-language trials by providing better estimation of within-speaker scatter matrix in WCCN and LDA techniques. However, neither language normalization nor the state-of-the-art PLDA algorithm is capable of modeling language variability on a dataset with insufficient multilingual utterances for each speaker, resulting in a poor performance in cross-language trial condition. This study is an extension to our initial developments of a language-independent PLDA training algorithm which aimed at reducing the effect of language as a source of variability on the performance of speaker recognition. We will provide a thorough analysis of how the proposed approach can utilize multilingual training data from bilingual speakers to robustly compensate for the effect of languages. Evaluated on multilingual trial condition, the proposed solution demonstrated over 10% EER and 13% minimum DCF relative improvement on NIST 2008 speaker recognition evaluation as well as 12.4% EER and 23% minimum DCF on PRISM evaluation set over the baseline system while also providing improvement in other trial conditions.

In this paper, we apply x-vectors to the task of spoken language recognition. This framework consists of a deep neural network that maps sequences of speech features to fixed-dimensional embeddings, called x-vectors. Longterm language characteristics are captured in the network by a temporal pooling layer that aggregates information across time. Once extracted, x-vectors utilize the same classification technology developed for i-vectors. In the 2017 NIST language recognition evaluation, x-vectors achieved excellent results and outperformed our state-ofthe-art i-vector systems. In the post-evaluation analysis presented here, we experiment with several variations of the x-vector framework, and find that the best performing system uses multilingual bottleneck features, data augmentation, and a discriminative Gaussian classifier.

in this paper we implement state of the art factor analysis based methods and fused their scores to gain a channel robust speaker recognition system. These two methods are joint factor analysis (JFA) and i-Vector which define low-dimensional speaker and channel dependent spaces. For score fusion we propose a simple weight computation without training step. We experiment our method on two conditions; 1) in channel matched training and test channel (telephone in training phase/telephone in test phase) task and 2) the channel mismatched condition (telephone training phase/microphone, GSM and VOIP in test phase) task. Our strategies outperform a state-of-the-art GMM-UBM based system. We obtained more than 4% absolute EER improvement for both channel dependent and channel independent condition compared to the standard GMM-UBM based method. Simulation also results that the combined i-Vector and JFA based system give better performance than all implemented method.

Speaker de-identification approaches must accomplish three main goals: universality, naturalness and reversibility. The main drawback of the traditional approach to speaker de-identification using voice conversion techniques is its lack of universality, since a parallel corpus between the input and target speakers is necessary to train the conversion parameters. It is possible to make use of a synthetic target to overcome this issue, but this harms the naturalness of the resulting de-identified speech. Hence, a technique is proposed in this paper in which a pool of pre-trained transformations between a set of speakers is used as follows: given a new user to de-identify, its most similar speaker in this set of speakers is chosen as the source speaker, and the speaker that is the most dissimilar to the source speaker is chosen as the target speaker. Speaker similarity is measured using the i-vector paradigm, which is usually employed as an objective measure of speaker de-identification performance, leading to a system with high de-identification accuracy. The transformation method is based on frequency warping and amplitude scaling, in order to obtain natural sounding speech while masking the identity of the speaker. In addition, compared to other voice conversion approaches, the proposed method is easily reversible. Experiments were conducted on Albayzin database, and performance was evaluated in terms of objective and subjective measures. These results showed a high success when de-identifying speech, as well as a great naturalness of the transformed voices. In addition, when making the transformation parameters available to a trusted holder, it is possible to invert the de-identification procedure, hence recovering the original speaker identity. The computational cost of the proposed approach is small, making it possible to produce de-identified speech in real-time with a high level of naturalness.

This paper focuses on the problem of ensemble classification for text-independent speaker verification. Ensemble classification is an efficient method to improve the performance of the classification system. This method gains the advantage of a set of expert classifiers. A speaker verification system gets an input utterance and an identity claim, then verifies the claim in terms of a matching score. This score determines the resemblance of the input utterance and preenrolled target speakers. Since there is a variety of information in a speech signal, state-of-the-art speaker verification systems use a set of complementary classifiers to provide a reliable decision about the verification. Such a system receives some scores as input and takes a binary decision: accept or reject the claimed identity. Most of the recent studies on the classifier fusion for speaker verification used a weighted linear combination of the base classifiers. The corresponding weights are estimated using logistic regression. Additional researches have been performed on ensemble classification by adding different regularization terms to the logistic regression formulae. However, there are missing points in this type of ensemble classification, which are the correlation of the base classifiers and the superiority of some base classifiers for each test instance. We address both problems, by an instance based classifier ensemble selection and weight determination method. Our extensive studies on NIST 2004 speaker recognition evaluation (SRE) corpus in terms of EER, minDCF and minCLLR show the effectiveness of the proposed method.

In recent years, there have been significant advances in the field of speaker recognition that has resulted in very robust recognition systems. The primary focus of many recent developments have shifted to the problem of recognizing speakers in adverse conditions, e.g in the presence of noise/reverberation. In this paper, we present the UMD-JHU speaker recognition system applied on the NIST 2010 SRE task. The novel aspects of our systems are: 1) Improved performance on trials involving different vocal effort via the use of linearscale features; 2) Expected improved recognition performance in the presence of reverberation and noise via the use of frequency domain perceptual linear predictor and cortical features; 3) A new discriminative kernel partial least squares (KPLS) framework that complements state-of-the-art back-end systems JFA and PLDA to aid in better overall recognition; and 4) Acceleration of JFA, PLDA and KPLS back-ends via distributed computing. The individual components of the system and the fused system are compared against a baseline JFA system and results reported by SRI and MIT-LL on SRE2010.

I-vectors are concise representations of speaker characteristics. Recent progress in i-vectors related research has utilized their ability to capture speaker and channel variability to develop efficient automatic speaker verification (ASV) systems. Inter-speaker relationships in the i-vector space are nonlinear. Accomplishing effective speaker verification requires a good modeling of these non-linearities and can be cast as a machine learning problem. Kernel partial least squares (KPLS) can be used for discriminative training in the i-vector space. However, this framework suffers from training data imbalance and asymmetric scoring. We use "one shot similarity scoring" (OSS) to address this. The resulting ASV system (OSS-KPLS) is tested across several conditions of the NIST SRE 2010 extended core data set and compared against state-of-the-art systems: Joint Factor Analysis (JFA), Probabilistic Linear Discriminant Analysis (PLDA), and Cosine Distance Scoring (CDS) classifiers. Improvements are shown.

In this paper, we propose a speaker-verification system based on maximum likelihood linear regression (MLLR) super-vectors, for which speakers are characterized by m-vectors. These vectors are obtained by a uniform segmentation of the speaker MLLR super-vector using an overlapped sliding window. We consider three approaches for MLLR transformation, based on the conventional 1-best automatic transcription, on the lattice word transcription, or on a simple global universal background model (UBM). Session variability compensation is performed in a post-processing module with probabilistic linear discriminant analysis (PLDA) or the eigen factor radial (EFR). Alternatively, we propose a cascade post-processing for the MLLR super-vector based speaker-verification system. In this case, the m-vectors or MLLR super-vectors are first projected onto a lower-dimensional vector space generated by linear discriminant analysis (LDA). Next, PLDA session variability compensation and scoring is applied to the reduced-dimensional vectors. This approach combines the advantages of both techniques and makes the estimation of PLDA parameters easier. Experimental results on telephone conversations of the NIST 2008 and 2010 speaker recognition evaluation (SRE) indicate that the proposed m-vector system performs significantly better than the conventional system based on the full MLLR super-vectors. Cascade post-processing further reduces the error rate in all cases. Finally, we present the results of fusion with a standard i-vector system in the feature, as well as in the score domain, demonstrating that the m-vector system is both competitive and complementary with it.

Over several decades, speaker recognition performance has steadily improved for applications using telephone speech. A big part of this improvement has been the availability of large quantities of speaker-labeled data from telephone recordings. For new data applications, such as audio from room microphones, we would like to effectively use existing telephone data to build systems with high accuracy while maintaining good performance on existing telephone tasks. In this paper we compare and combine approaches to compensate models parameters and features for this purpose. For model adaptation we explore MAP adaptation of hyper-parameters and for feature compensation we examine the use of denoising DNNs. On a multi-room, multi-microphone speaker recognition experiment we show a reduction of 61% in EER with a combination of these approaches while slightly improving performance on telephone data.

Speaker recognition systems frequently use GMM-MAP method for modeling speakers. This method represents a speaker using a Gaussian mixture. However in this mixture not all the Gaussian components are truly representative of the speaker. In order to remove the model redundancy, this work proposes a Gaussian selection method to achieve a new GMM model only with the more representative Gaussian components. Speaker verification experiments applying the proposal show a similar performance to baseline; however the speaker models have a reduction of 80 % regarding the speaker model used for baseline. The application of this Gaussian selection method in real or embedded speaker verification systems could be very useful for reducing computational and memory cost.

This paper describes the speaker verification (SV) system submitted to the NIST 2016 speaker recognition evaluation (SRE) challenge by Indian Institute of Technology Guwahati (IITG) under the fixed training condition task. Various SV systems are developed following the idea-level collaboration with two other Indian institutions. Unlike the previous SREs, this time the focus was on developing SV system using non-target language speech data and a small amount unlabeled data from target language/dialects. For addressing these novel challenges, we tried exploring the fusion of systems created using different features, data conditioning, and classifiers. On NIST 2016 SRE evaluation data, the presented fused system resulted in actual detection cost function (actDCF) and equal error rate (EER) of 0.81 and 12.91%, respectively. Post-evaluation, we explored a recently proposed pairwise support vector machine classifier and applied adaptive S-norm to the decision scores before fusion. With these changes, the final system achieves the actDCF and EER of 0.67 and 11.63%, respectively.

I4U is a joint entry of nine research Institutes and Universities across 4 continents to NIST SRE 2012. It started with a brief discussion during the Odyssey 2012 workshop in Singapore. An online discussion group was soon set up, providing a discussion platform for different issues surrounding NIST SRE'12. Noisy test segments, uneven multi-session training, variable enrollment duration, and the issue of open-set identification were actively discussed leading to various solutions integrated to the I4U submission. The joint submission and several of its 17 subsystems were among top-performing systems. We summarize the lessons learnt from this large-scale effort.

Speaker Identification process is to identify a particular vocal cord from a set of existing speakers. In the speaker identification processes, unknown speaker voice sample targets each of the existing speakers present in the system and gives a predication. The predication may be more than one existing known speaker voice and is very close to the unknown speaker voice. The model is a Gaussian mixture model built by the extracted acoustic feature vectors from voice. The i-vector based dimension compression mapping function of the channel depended speaker, and super vector give better predicted scores according to cosine distance scoring associated with the order pair of speakers. In the order pair, the first coordinate is the unknown speaker i.e. test speaker, and the second coordinates is the existing known speaker i.e. target speaker. This paper presents the enhancement of the prediction based on i- vector in compare to the normalized set of predicted score. In the simulation, know...

Speaker verification in real-world applications sometimes deals with limited duration of enrollment and/or test data. MFCC-based i-vector systems have defined the state-of-the-art for speaker verification, but it is well known that they are less effective with short utterances. To address this issue, we propose a method to leverage the speaker specificity and stationarity of subglottal acoustics. First, we present a deep neural network (DNN) based approach to estimate subglottal features from speech signals. The approach involves training a DNN-regression model that maps the log filter-bank coefficients of a given speech signal to those of its corresponding subglottal signal. Cross-validation experiments on the WashU-UCLA corpus (which contains parallel recordings of speech and subglottal acoustics) show the effectiveness of our DNN-based estimation algorithm. The average correlation coefficient between the actual and estimated subglottal filter-bank coefficients is 0.9. A scorelevel fusion of MFCC and subglottal-feature systems in the ivector PLDA framework yields statistically-significant improvements over the MFCC-only baseline. On the NIST SRE 08 truncated 10sec-10sec and 5sec-5sec core evaluation tasks, the relative reduction in equal error rate ranges between 6 and 14% for the conditions tested with both microphone and telephone speech.

— This document briefly describes the systems submitted by the Center for Robust Speech Systems (CRSS) from The University of Texas at Dallas (UTD) for the 2012 NIST Speaker Recognition Evaluation. We developed a state-of-the-art i-vector based speaker recognition system [1]. Probabilistic linear discriminant analysis (PLDA) [2] along with several other backends are used for channel/noise compensation. Given that the emphasis of the NIST SRE-2012 is on noisy and short duration test conditions, in our system development we focused on: (1) novel robust acoustic features, (2) new feature normalization schemes, (3) various back-end strategies for multiple session enrollment. I.

Factor analysis based channel mismatch compensation methods for speaker recognition are based on the assumption that speaker/utterance dependent Gaussian Mixture Model (GMM) mean super-vectors can be constrained to reside in a lower dimensional subspace. This approach does not consider the fact that conventional acoustic feature vectors also reside in a lower dimensional manifold of the feature space, when feature covariance matrices contain close to zero eigenvalues. In this study, based on observations of the covariance structure of acoustic features, we propose a factor analysis modeling scheme in the acoustic feature space instead of the super-vector space and derive a mixture dependent feature transformation. We demonstrate how this single linear transformation performs feature dimensionality reduction, de-correlation, normalization and enhancement, at once. The proposed transformation is shown to be closely related to signal subspace based speech enhancement schemes. In contrast to traditional front-end mixture dependent feature transformations, where feature alignment is performed using the highest scoring mixture, the proposed transformation is integrated within the speaker recognition system using a probabilistic feature alignment technique, which nullifies the need for regenerating the features/retraining the Universal Background Model (UBM). Incorporating the proposed method with a state-of-the-art i-vector and Gaussian Probabilistic Linear Discriminant Analysis (PLDA) framework, we perform evaluations on National Institute of Science and Technology (NIST) Speaker Recognition Evaluation (SRE) 2010 core telephone and microphone tasks. The experimental results demonstrate the superiority of the proposed scheme compared to both full-covariance and diagonal covariance UBM based systems. Simple equal-weight fusion of baseline and proposed systems also yield significant performance gains. Index Terms-Acoustic feature enhancement, factor analysis, probabilistic principal component analysis, speaker verification. I. INTRODUCTION M ISMATCH between training and test conditions represent one of the most challenging problems facing speaker recognition researchers today. There can be consid

This paper describes the speaker verification (SV) system submitted to the NIST 2016 speaker recognition evaluation (SRE) challenge by Indian Institute of Technology Guwahati (IITG) under the fixed training condition task. Various SV systems are developed following the idea-level collaboration with two other Indian institutions. Unlike the previous SREs, this time the focus was on developing SV system using non-target language speech data and a small amount unlabeled data from target language/dialects. For addressing these novel challenges, we tried exploring the fusion of systems created using different features, data conditioning, and classifiers. On NIST 2016 SRE evaluation data, the presented fused system resulted in actual detection cost function (actDCF) and equal error rate (EER) of 0.81 and 12.91%, respectively. Post-evaluation, we explored a recently proposed pairwise support vector machine classifier and applied adaptive S-norm to the decision scores before fusion. With these changes, the final system achieves the actDCF and EER of 0.67 and 11.63%, respectively.

In recent years, there have been significant advances in the field of speaker recognition that has resulted in very robust recognition systems. The primary focus of many recent developments have shifted to the problem of recognizing speakers in adverse conditions, e.g in the presence of noise/reverberation. In this paper, we present the UMD-JHU speaker recognition system applied on the NIST 2010 SRE task. The novel aspects of our systems are: 1) Improved performance on trials involving different vocal effort via the use of linearscale features; 2) Expected improved recognition performance in the presence of reverberation and noise via the use of frequency domain perceptual linear predictor and cortical features; 3) A new discriminative kernel partial least squares (KPLS) framework that complements state-of-the-art back-end systems JFA and PLDA to aid in better overall recognition; and 4) Acceleration of JFA, PLDA and KPLS back-ends via distributed computing. The individual components of the system and the fused system are compared against a baseline JFA system and results reported by SRI and MIT-LL on SRE2010.

This paper evaluates the performance of the twelve primary systems submitted to the evaluation on speaker verification in the context of a mobile environment using the MOBIO database. The mobile environment provides a challenging and realistic test-bed for current state-of-the-art speaker verification techniques. Results in terms of equal error rate (EER), half total error rate (HTER) and detection error trade-off (DET) confirm that the best performing systems are based on total variability modeling, and are the fusion of several subsystems. Nevertheless, the good old UBM-GMM based systems are still competitive. The results also show that the use of additional data for training as well as gender-dependent features can be helpful.

Current state-of-the-art speaker verification (SV) systems are known to be strongly affected by unexpected variability presented during testing, such as environmental noise or changes in vocal effort. In this work, we analyze and evaluate articulatory information of the tongue's movement as a means to improve the performance of speaker verification systems. We use a Spanish database, where besides the speech signals, we also include articulatory information that was acquired with an ultrasound system. Two groups of features are proposed to represent the articulatory information, and the obtained performance is compared to an SV system trained only with acoustic information. Our results show that the proposed features contain highly discriminative information, and they are related to speaker identity; furthermore, these features can be used to complement and improve existing systems by combining such information with cepstral coefficients at the feature level.