Editing Spoofing attack (section)

== Voice spoofing ==
Information technology plays an increasingly large role in today's world, and different authentication methods are used for restricting access to informational resources, including voice biometrics. Examples of using [[speaker recognition]] systems include internet banking systems, customer identification during a call to a call center, as well as passive identification of a possible criminal using a preset "blacklist".<ref>{{Cite conference|last1=Shchemelinin|first1=Vadim|last2=Topchina|first2=Mariia|last3=Simonchik|first3=Konstantin|date=2014|editor-last=Ronzhin|editor-first=Andrey|editor2-last=Potapova|editor2-first=Rodmonga|editor3-last=Delic|editor3-first=Vlado|title=Vulnerability of Voice Verification Systems to Spoofing Attacks by TTS Voices Based on Automatically Labeled Telephone Speech|url=https://link.springer.com/chapter/10.1007/978-3-319-11581-8_59|conference=International Conference on Speech and Computer|series=Lecture Notes in Computer Science|volume=8773|language=en|location=Cham|publisher=Springer International Publishing|pages=475–481|doi=10.1007/978-3-319-11581-8_59|isbn=978-3-319-11581-8|url-access=subscription}}</ref>

Technologies related to the synthesis and modeling of speech are developing very quickly, allowing one to create voice recordings almost indistinguishable from real ones. Such services are called [[Speech synthesis|Text-to-Speech]] (TTS) or [[Neural Style Transfer|Style transfer]] services. The first one aimed at creating a new person. The second one aimed at identifies as another in voice identification systems.

A large number of scientists are busy developing algorithms that would be able to distinguish the synthesized voice of the machine from the real one. On the other hand, these algorithms need to be thoroughly tested to make sure that the system really works.<ref>{{Cite book|last1=Sinitca|first1=Aleksandr M.|last2=Efimchik|first2=Nikita V.|last3=Shalugin|first3=Evgeniy D.|last4=Toropov|first4=Vladimir A.|last5=Simonchik|first5=Konstantin|title=2020 IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering (EIConRus) |chapter=Voice Antispoofing System Vulnerabilities Research |date=January 2020|chapter-url=https://ieeexplore.ieee.org/document/9039393|location=St. Petersburg and Moscow, Russia|publisher=IEEE|pages=505–508|doi=10.1109/EIConRus49466.2020.9039393|isbn=978-1-7281-5761-0|s2cid=214595791}}</ref> However, an early study has shown that feature design and masking augmentation have a significant impact on the ability to detect spoofed voice.<ref>{{cite journal |last1=Cohen |first1=Ariel |last2=Rimon |first2=Inbal |last3=Aflalo |first3=Eran |last4=Permuter |first4=Haim H. |title=A study on data augmentation in voice anti-spoofing |journal=Speech Communication |date=June 2022 |volume=141 |pages=56–67 |doi=10.1016/j.specom.2022.04.005|arxiv=2110.10491 |s2cid=239050551 }}</ref>