Voice Biometrics Technology, Trust and Security edited by Carmen García-Mateo, Gérard Chollet

Format:

Book

Contributor:

Chollet, Gérard.

Series:

Security Series

IET Security Series 12

Language:

English

Physical Description:

1 recurso en línea (898 p.) il., gráf. y tablas

Edition:

1st ed.

Other Title:

Voice Biometrics

Place of Publication:

London, UK Institution of Engineering & Technology 2021

Summary:

La biometría de voz se está implementando a nivel mundial en aplicaciones a gran escala como banca remota, servicios electrónicos gubernamentales, transporte y acceso de seguridad a edificios, vehículos autónomos y atención médica. Se han integrado en numerosas aplicaciones, a menudo junto con métodos biométricos faciales e inteligencia artificial. Los productos y soluciones de biometría de voz deben cumplir con tres requisitos clave para el éxito de su implementación: deben ser altamente confiables con respecto a la protección de la privacidad; fácil de usar y siempre disponible.

Contents:

Intro

Halftitle Page

Series Page

Title Page

Copyright

Contents

List of figures

List of tables

Short biographies of the editors and authors

Preface to Voice Biometrics

About the editors

1 Introduction

Chapter 2 - Fundamentals of voice biometrics: classical and machine learning approaches

Chapter 3 - Voice biometrics: attacker's perspective

Chapter 4 - Voice biometrics: privacy in paralinguistic and extralinguistic tasks for health applications

Chapter 5 - Voice privacy in biometrics: speaker de-identification

Chapter 6 - Performance evaluation of voice biometrics solutions

Chapter 7 - Voice biometrics: how the technology is standardized

Chapter 8 - Voice biometrics: perspective from the industry

Chapter 9 - Joining forces of voice and facial biometrics: a case study in the scope of NIST SRE'19

Chapter 10 - Voice biometrics: future trends and challenges ahead

2 Fundamentals of voice biometrics: classical and machine learning approaches

2.1 Introduction to speaker recognition systems

2.2 Metrics for system performance evaluation

2.2.1 ROC, DET and EER

2.2.2 Detection cost function

2.3 Text-independent speaker recognition

2.3.1 Classical acoustic approaches: GMM-UBM, i-vector and PLDA

2.3.2 DNN approaches

2.3.2.1 Basic concepts of neural networks

2.3.2.2 Some applications of DNNs to speech processing

2.3.3 DNNs for speaker recognition

2.4 Text-dependent speaker recognition

2.4.1 Classification of systems and techniques

2.4.2 Databases and benchmarks

2.5 Calibration of speaker recognition scores

2.5.1 Motivation: why to calibrate?

2.5.2 What is calibration?

2.5.3 Score-to-LR computation methods

2.5.3.1 Generative calibration models: fitting distributions to scores.

2.5.3.2 Discriminative calibration models: transforming scores into LR values to optimize a cost function

2.5.4 Performance measurement of score-to-LR methods

References

3 Voice biometrics: attacker's perspective

Abstract

3.1 Introduction

3.2 Direct attacks

3.2.1 Spoofing attacks

3.2.2 Black box hardware attacks

3.2.3 Black box adversarial attacks

3.3 Indirect attacks

3.3.1 Attacks on corpora

3.3.2 Gray box hardware attacks

3.3.3 Gray box and white box adversarial attacks

3.4 Technological challenges

3.4.1 Extracting prosodic information

3.4.2 Enrolled users with malicious intent

3.4.3 Number of trials permitted on the ASV

3.4.4 Minuteness of the perturbation in adversarial attacks

3.4.5 Privacy preservation of speech and voice privacy

3.5 Conclusions and future work

Acknowledgments

4 Voice biometrics: privacy in paralinguistic and extralinguistic tasks for health applications

4.1 Introduction

4.2 Paralinguistic and extralinguistic tasks

4.2.1 Speech-affecting diseases

4.2.2 Methods

4.3 Cryptographic primitives and MPC for PPML

4.3.1 Homomorphic encryption

4.3.2 Oblivious transfer

4.3.3 Secure Multiparty Computation

4.3.3.1 Yao's GCs protocol

4.3.3.2 Secret sharing

4.3.3.3 Security models

4.3.4 Distance-preserving hashing techniques

Secure binary embeddings

Secure modular hashing

4.4 PPML for paralinguistic and extralinguistic tasks

4.4.1 PPML for non-health-related tasks

4.4.2 PPML for health-related tasks

4.4.3 Private SVM+RBF for health-related tasks

4.4.3.1 Private RBF computation

4.4.3.2 Private SVM computation

4.4.3.3 Experimental setup

4.4.3.4 Model training and parameters

4.4.3.5 Private SVM implementation details.

4.4.3.6 Classification results

4.4.3.7 Security and computational performance

4.5 Conclusions

Acknowledgements

5 Voice privacy in biometrics: speaker de-identification

5.1 Introduction

5.2 How to evaluate speaker de-identification?

5.2.1 Subjective measures

5.2.2 Objective measures

5.3 Speaker de-identification techniques

5.3.1 Codebook mapping

5.3.2 Gaussian mixture model

5.3.3 Frequency warping

5.3.4 Deep learning techniques

5.4 Experiment definition

5.4.1 Piecewise definition of transformation functions

5.4.2 Pretrained transformation functions

5.4.3 De-identification based on DNNs

5.4.4 De-identification based on generative adversarial networks

5.5 Evaluation corpora

5.5.1 Evaluation metrics

5.6 Results and analysis

5.7 Conclusion

6 Performance evaluation of voice biometrics solutions

6.1 Introduction

6.2 Evaluating methods or technology

6.2.1 Existing benchmarking evaluations

6.2.2 Evaluation criteria

6.2.2.1 Evaluating a system producing hard decisions

6.2.2.2 Evaluating the goodness of verification scores

6.2.3 Statistical significance

6.2.4 Specific evaluation aspects

6.2.5 Evaluating related technologies

6.3 Bias in testing

6.4 Summary and propositions

7 Voice biometrics: How the technology is standardized

7.1 Introduction

7.2 Biometrics standardization within ISO/IEC

7.2.1 Generalized system design

7.2.2 Harmonized biometric vocabulary

7.2.3 Performance testing and reporting

7.2.4 Presentation attack detection

7.2.5 Biometric information protection

7.3 Data interchange formats for passports and beyond

7.3.1 Motivation and background on encoding biometric data.

7.3.2 Data interchange standard ISO/IEC 19794

7.3.3 Format structure

7.3.4 ISO/IEC 19794 Part 13: voice data

7.4 Discussion: de facto and ISO/IEC standards

7.4.1 On the general system design

7.4.2 Gap analysis: performance testing and reporting

7.4.3 Regarding implementations and data interchange formats

7.5 Conclusion

8 Voice biometrics: perspective from the industry

8.1 Automated password reset: an example of a commercial application using voice biometrics

8.1.1 Overview

8.1.2 Introduction

8.1.3 System architecture

8.1.4 Voice biometric system

8.1.5 Summary

8.2 Testing of commercial voice biometric systems

8.2.1 Introduction

8.2.1.1 Biometric testing

8.2.2 User analysis

8.2.3 Summary

8.3 Forensic speaker recognition

8.3.1 Introduction

8.3.2 Forensic speaker recognition and the strength of evidence

8.3.3 The forensic expert's workflow

8.3.4 Technical challenges

8.3.4.1 Improving interpretability of scores

8.3.4.2 Score normalization

8.3.4.3 Score calibration

8.3.4.4 Condition adaptation

8.3.4.5 Dealing with multi-speaker recordings

8.3.5 Training-communication between system developers and end-users

8.3.6 Conclusions

9 Joining forces of voice and facial biometrics: a case study in the scope of NIST SRE'19

9.1 Introduction to the NIST SRE'19 challenge

9.1.1 The SRE'19 CTS challenge

9.1.2 The SRE'19 multimedia challenge

9.1.3 SRE'19 evaluation metrics

9.2 TSP speaker verification system for the SRE'19 evaluation

9.2.1 A brief review of state of the art in speaker verification

9.2.2 TSP speaker verification common pipeline for the SRE'19 CTS and multimedia challenges

TDNN

E-TDNN.

9.2.3 TSP speaker verification system for the SRE'19 CTS challenge

9.2.4 TSP speaker verification system for the SRE'19 multimedia challenge

9.2.5 Results for TSP speaker verification systems on the SRE'19 CTS and multimedia challenges

9.2.6 Conclusions

9.3 TSP face recognition system for SRE'19

9.3.1 Survey of face recognition systems

9.3.2 TSP face recognition system pipeline

9.3.3 Databases used in the TSP face recognition system

9.3.4 Face preprocessing

9.3.5 Embedding extractor

Initial version of the DNN architecture

Final version of the DNN architecture

9.3.6 Conclusions

9.4 Audiovisual biometric system for the SRE'19 multimedia challenge

9.5 Conclusions and perspectives

10 Voice biometrics: future trends and challenges ahead

10.1 Applications

10.2 Privacy and security

10.3 Research

Index

Back Cover.

Notes:

Índice

Description based on publisher supplied metadata and other sources.

ISBN:

9781785619014

1785619012

OCLC:

1264473157