The dawn of a text-dependent society: deepfakes as a threat to speech verification systems

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F22%3APU142956" target="_blank" >RIV/00216305:26230/22:PU142956 - isvavai.cz</a>

Výsledek na webu

<a href="https://dl.acm.org/doi/10.1145/3477314.3507013" target="_blank" >https://dl.acm.org/doi/10.1145/3477314.3507013</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1145/3477314.3507013" target="_blank" >10.1145/3477314.3507013</a>

Jazyk výsledku

angličtina

Název v původním jazyce

The dawn of a text-dependent society: deepfakes as a threat to speech verification systems

Popis výsledku v původním jazyce

We are already aware that deepfakes pose threats to humankind. Nowadays, mostly as fake news or disinformation; however, there are still unexplored areas such as using deepfakes to spoof voice verification. We present a real-world use case for spoofing voice authentication in a customer care call center. Based on this scenario, we evaluate the feasibility of attacking such a system and create an attacker profile. For this purpose, we examine three available speech synthesis tools and discuss their usability. We use these tools and acquired knowledge to generate a dataset including deepfake speech and assess the resilience of voice biometrics systems against deepfakes. We prove that voice biometrics systems are indeed vulnerable to deepfake powered attacks. The most significant outcome is the proposal of text-dependent verification as a novel countermeasure for presented attacks. Text-dependent verification provides higher security than text-independent verification and can be used today as the simplest protection method against deepfakes.

Název v anglickém jazyce

The dawn of a text-dependent society: deepfakes as a threat to speech verification systems

Popis výsledku anglicky

We are already aware that deepfakes pose threats to humankind. Nowadays, mostly as fake news or disinformation; however, there are still unexplored areas such as using deepfakes to spoof voice verification. We present a real-world use case for spoofing voice authentication in a customer care call center. Based on this scenario, we evaluate the feasibility of attacking such a system and create an attacker profile. For this purpose, we examine three available speech synthesis tools and discuss their usability. We use these tools and acquired knowledge to generate a dataset including deepfake speech and assess the resilience of voice biometrics systems against deepfakes. We prove that voice biometrics systems are indeed vulnerable to deepfake powered attacks. The most significant outcome is the proposal of text-dependent verification as a novel countermeasure for presented attacks. Text-dependent verification provides higher security than text-independent verification and can be used today as the simplest protection method against deepfakes.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2022

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

SAC '22: Proceedings of the 37th ACM/SIGAPP Symposium on Applied Computing

ISBN

978-1-4503-8713-2

ISSN

—

e-ISSN

—

Počet stran výsledku

10

Strana od-do

1646-1655

Název nakladatele

Association for Computing Machinery

Místo vydání

New York, NY

Místo konání akce

Brno

Datum konání akce

25. 4. 2022

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—