García-Mateo Carmen - Voice Biometrics

IET Book Series in Advances in Biometrics Call for authors

Book Series Editor:Michael Fairhurst, University of Kent, UK

This Book Series provides the foundation on which a valuable library of reference volumes on the topic of Biometrics is build. Iris and Periocular Biometric Recognition, Mobile Biometrics and User-centric Privacyand Security in Biometrics are the first published volumes in the Series, with further titles currently being commissioned. Proposals for coherently integrated, multi-author edited contributions are welcome for consideration. Please email your proposal to the Book Series Editor, Professor Michael Fairhurst, at: .

Published Titles in This Series:

Published 2018 Drahansk Martin (ed) / Hand-Based Biometrics: Methods and technology

Published 2017 Guo and Wechsler (Eds) / Mobile Biometrics

Published 2017 Rathgeb and Bush (eds) / Iris and Periocular Biometric Recognition.

Published 2017 Vielhauer / Privacy and Security in Biometrics

Published 2013 Fairhurst / Age Factors in Biometrics Processing

Edited by
Carmen Garca-Mateo and Grard Chollet

The Institution of Engineering and Technology

Published by The Institution of Engineering and Technology, London, United Kingdom

The Institution of Engineering and Technology is registered as a Charity in England & Wales (no. 211014) and Scotland (no. SC038698).

The Institution of Engineering and Technology 2021

First published 2021

This publication is copyright under the Berne Convention and the Universal Copyright Convention. All rights reserved. Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may be reproduced, stored or transmitted, in any form or by any means, only with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms of licences issued by the Copyright Licensing Agency. Enquiries concerning reproduction outside those terms should be sent to the publisher at the undermentioned address:

The Institution of Engineering and Technology
Michael Faraday House
Six Hills Way, Stevenage
Herts, SG1 2AY, United Kingdom

www.theiet.org

While the authors and publisher believe that the information and guidance given in this work are correct, all parties must rely upon their own skill and judgement when making use of them. Neither the author nor publisher assumes any liability to anyone for any loss or damage caused by any error or omission in the work, whether such an error or omission is the result of negligence or any other cause. Any and all such liability is disclaimed.

The moral rights of the author to be identified as author of this work have been asserted by him in accordance with the Copyright, Designs and Patents Act 1988.

British Library Cataloguing in Publication Data
A catalogue record for this product is available from the British Library

ISBN 978-1-78561-900-7 (hardback)
ISBN 978-1-78561-901-4 (PDF)

Typeset in India by Exeter Premedia Services Private Limited
Printed in the UK by CPI Group (UK) Ltd, Croydon

This figure represents a standard feed-forward DNN architecture, with an input layer, a given number of hidden layers and the output layer. Each hidden unit applies a transformation g (usually non-linear function) to the weighted sum of the inputs to that given unit according to the following expression: , where W and b are the parameters of the DNN (weights matrices and bias vectors, respectively). Finally, the output layer provides the score vector according to the target task (e.g., for the case of classification, the probability of an input vector belong to each class)

This figure shows a graphical representation of a cepstral (upper part) versus bottleneck (bottom part) feature-based speaker recognition system. For the bottleneck (BN) feature-based approach, sometimes, MFCCs are concatenated with the bottleneck features before the i-vector modeling

This is a representation of the generic scheme of a DNN used as embedding extractor

Direct vs. indirect attacks. After []

Classification of various attacks on an ASV system

Motivation for attacks on ASV

SSD system for ASV system

Using ASV to attack the victim ASV

Privacy-preservation for speech corpus. After []

(a) Attack using target selection but without voice privacy system, and (b) attack using target selection with voice privacy system

Example of a privacy-preserving machine learning framework

Fifth degree Chebyshev polynomial approximation ()

Piecewise linear approximation of a FW

Diagram of the proposed speaker selection strategy for voice conversion. The dots represent the different speakers, and those inside the blue ellipse are training speakers that share a parallel corpus

DET curves of the different systems on the VCTK corpus

DET curves of the different systems on the VCC 2016 corpus

General design of a biometric system; source and shape semantics: ISO/IEC 39794-1 []

Overview on attack points in biometric systems with depicted scope of ISO/IEC 30107-1 []

General biometric system composition with PAD; source and shape semantics: ISO/IEC 30107-1 []

Typical architecture for a commercial automated password reset system

Standard deviation of error rate estimate as a function of N. FA: false accept; STD, standard deviation

Histogram of false accept rates across 10000 population with global threshold

95% Confidence interval for user false accept rate using global and per-user thresholds. FA: false accept

Histogram of false accept rates across 10000 population with per-user thresholds

An example of LR estimation given the same-speaker (H0) and different-speaker (H1) hypotheses and an evidential speaker recognition score E

Block diagram of TSPs common speaker verification systems for NIST 2019 CTS and multimedia challenges

DET performance curves on the development set of SRE19 CTS database for system (1) based on TDNN, system (2) based on E-TDNN, and the fusion of (1) and (2)

Block diagram of the TSP face recognition system for SRE19 multimedia challenge

The Multi-PIE 68 points markup [] used for face landmark annotation

Example of the preprocessing of an image from LFW using eyes and nose positions

Examples from the alignment of images from the ATSIP2018 database with good acquisition conditions, i.e., frontal face, good illumination

Examples from the alignment of images from the ATSIP2018 database with bad acquisition conditions, i.e., face turned to a great degree

DET curve performance of the submitted face recognition system on the dev and test partitions of the multimedia challenge