Channel Compensation for Speaker Recognition using MAP Adapted PLDA and Denoising DNNs

Richardson, F. S.; Nemsick, Brian E.; Reynolds, Douglas A.

doi:10.21437/odyssey.2016-32

Cited by 11 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Both theoretical derivation and experiments conducted on the NIST SRE10 core condition demonstrate that: 1) the parameterization of JB enables it to learn the intrinsic dimensionality of the identify subspace, which can reduce the system complexity without performance degradation; 2) Hidden variables selection of JB makes EM iterations converge faster with better parameter estimation; 3) The EM with exact statistics performs better than with approximated statistics. For future work, it is interesting to apply data domain adaption [5] and feature compensation [14,15] and nearestneighbor discriminant analysis (NDA) [16][17] that have been successfully applied to PLDA to JB to further improve performance.…”

Section: Discussionmentioning

confidence: 99%

Joint Bayesian Gaussian Discriminant Analysis for speaker verification

Wang

2017

2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

State-of-the-art i-vector based speaker verification relies on variants of Probabilistic Linear Discriminant Analysis (PLDA) for discriminant analysis. We are mainly motivated by the recent work of the joint Bayesian (JB) method, which is originally proposed for discriminant analysis in face verification. We apply JB to speaker verification and make three contributions beyond the original JB. 1) In contrast to the EM iterations with approximated statistics in the original JB, the EM iterations with exact statistics are employed and give better performance. 2) We propose to do simultaneous diagonalization (SD) of the within-class and between-class covariance matrices to achieve efficient testing, which has broader application scope than the SVD-based efficient testing method in the original JB. 3) We scrutinize similarities and differences between various Gaussian PLDAs and JB, complementing the previous analysis of comparing JB only with Prince-Elder PLDA. Extensive experiments are conducted on NIST SRE10 core condition 5, empirically validating the superiority of JB with faster convergence rate and 9 − 13% EER reduction compared with state-of-the-art PLDA.

show abstract

Section: Discussionmentioning

confidence: 99%

Joint Bayesian Gaussian Discriminant Analysis for speaker verification

Wang

2017

2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

show abstract

“…The input to the DNN is a 21 frame window using a set of stacked 40-dimensional MFCC (as in the Section 3.3.1.2) and the target was the 40 dimensional feature vector at the center of the input window extracted from the clean data. The general setup for this system is very close to that described in [12]. A 7x1024 layer senone classifying DNN was trained on 300 hours of SWB with ~8K senone targets.…”

Section: Mit-ll Denoising Stats I-vector Systemmentioning

confidence: 99%

The MIT-LL, JHU and LRDE NIST 2016 Speaker Recognition Evaluation System

et al. 2017

Self Cite

View full text Add to dashboard Cite

In this paper, the NIST 2016 SRE system that resulted from the collaboration between MIT Lincoln Laboratory and the team at Johns Hopkins University is presented. The submissions for the 2016 evaluation consisted of three fixed condition submissions and a single system open condition submission. The primary submission on the fixed (and core) condition resulted in an actual DCF of .618. Details of the submissions are discussed along with some discussion and observations of the 2016 evaluation campaign.

show abstract

“…It was recently found that clean i-vectors can be partly restored by translating and rotating the noisy i-vectors, where the translation and rotation matrices are found by the Kabsch algorithm [28]. Instead of denosing the i-vectors, spectral features can also be denoised by DNNs [29] and denoising autoencoder [30] before i-vector extraction.…”

Section: Related Workmentioning

confidence: 99%

DNN-Driven Mixture of PLDA for Robust Speaker Verification

Mak

Chien

2017

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

The mismatch between enrollment and test utterances due to different types of variabilities is a great challenge in speaker verification. Based on the observation that the SNRlevel variability or channel-type variability causes heterogeneous clusters in i-vector space, this paper proposes to apply supervised learning to drive or guide the learning of PLDA mixture models. Specifically, a deep neural network (DNN) is trained to produce the posterior probabilities of different SNR levels or channel types given i-vectors as input. These posteriors then replace the posterior probabilities of indicator variables in the mixture of PLDA. The discriminative training causes the mixture model to perform more reasonable soft divisions of the i-vector space as compared to the conventional mixture of PLDA. During verification, given a test i-vector and a targetspeaker's i-vector, the marginal likelihood for the same-speaker hypothesis is obtained by summing the component likelihoods weighted by the component posteriors produced by the DNN, and likewise for the different-speaker hypothesis. Results based on NIST 2012 SRE demonstrate that the proposed scheme leads to better performance under more realistic situations where both training and test utterances cover a wide range of SNRs and different channel types. Unlike the previous SNR-dependent mixture of PLDA which only focuses on SNR mismatch, the proposed model is more general and is potentially applicable to addressing different types of variability in speech.

show abstract

Channel Compensation for Speaker Recognition using MAP Adapted PLDA and Denoising DNNs

Cited by 11 publications

References 15 publications

Joint Bayesian Gaussian Discriminant Analysis for speaker verification

Joint Bayesian Gaussian Discriminant Analysis for speaker verification

The MIT-LL, JHU and LRDE NIST 2016 Speaker Recognition Evaluation System

DNN-Driven Mixture of PLDA for Robust Speaker Verification

Contact Info

Product

Resources

About