Deep Quality-Informed Score Normalization for Privacy-Friendly Speaker Recognition in Unconstrained Environments

In scenarios that are ambitious to protect sensitive data in compliance with privacy regulations, conventional score normalization utilizing large proportions of speaker cohort data is not feasible for existing technology, since the entire cohort data would need to be stored on each mobile device. Hence, in this work we motivate score normalization utilizing deep neural networks. Considering unconstrained environments, a quality-informed scheme is proposed, normalizing scores depending on sample quality estimates in terms of completeness and signal degradation by noise. Utilizing the conventional PLDA score, comparison i-vectors, and corresponding quality vectors, we aim at mimicking cohort based score normalization optimizing the minCllr discrimination criterion. Examining the I4U data sets for the 2012 NIST SRE, an 8.7% relative gain is yielded in a pooled 55-condition scenario with a corresponding condition-averaged relative gain of 6.2% in terms of minCllr. Robustness analyses towards sensitivity regarding unseen conditions are conducted, i.e. when conditions comprising lower quality samples are not available during training.

[1]  Radford M. Neal.  Pattern Recognition and Machine Learning , 2007, Technometrics.

[2]  Jian Sun,et al.  Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[3]  Patrick Kenny,et al.  Front-End Factor Analysis for Speaker Verification , 2011, IEEE Transactions on Audio, Speech, and Language Processing.

[4]  Lukás Burget,et al.  A unified approach for audio characterization and its application to speaker recognition , 2012, Odyssey.

[5]  Driss Matrouf,et al.  Identify the Benefits of the Different Steps in an i-Vector Based Speaker Verification System , 2013, CIARP.

[6]  John H. L. Hansen,et al.  Duration mismatch compensation for i-vector based speaker recognition systems , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[7]  Douglas A. Reynolds,et al.  Speaker Verification Using Adapted Gaussian Mixture Models , 2000, Digit. Signal Process..

[8]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[9]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[10]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[11]  David A. van Leeuwen,et al.  Quality measures based calibration with duration and noise dependency for speaker recognition , 2015, Speech Commun..

[12]  Niko Brümmer,et al.  Application-independent evaluation of speaker detection , 2006, Comput. Speech Lang..

[13]  Christoph Busch,et al.  Analysis of mutual duration and noise effects in speaker recognition: benefits of condition-matched cohort selection in score normalization , 2015, INTERSPEECH.

[14]  Daniel Garcia-Romero,et al.  Analysis of i-vector Length Normalization in Speaker Recognition Systems , 2011, INTERSPEECH.

[15]  Pietro Laface,et al.  Generative pairwise models for speaker recognition , 2014, Odyssey.

[16]  Patrick Kenny,et al.  Joint Factor Analysis of Speaker and Session Variability: Theory and Algorithms , 2006 .

[17]  David A. van Leeuwen,et al.  Quality Measure Functions for Calibration of Speaker Recognition Systems in Various Duration Conditions , 2013, IEEE Transactions on Audio, Speech, and Language Processing.

[18]  Douglas E. Sturim,et al.  Speaker adaptive cohort selection for Tnorm in text-independent speaker verification , 2005, Proceedings. (ICASSP '05). IEEE International Conference on Acoustics, Speech, and Signal Processing, 2005..

[19]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[20]  Pietro Laface,et al.  Comparison of Speaker Recognition Approaches for Real Applications , 2011, INTERSPEECH.

[21]  Christoph Busch,et al.  Robustness of Quality-based Score Calibration of Speaker Recognition Systems with respect to low-SNR and short-duration conditions , 2016, Odyssey.

[22]  John H. L. Hansen,et al.  I4u submission to NIST SRE 2012: a large-scale collaborative effort for noise-robust speaker verification , 2013, INTERSPEECH.