A Protection Scheme With Speech Processing Against Audio Adversarial Examples

Machine learning technologies have improved the accuracy of speech recognition systems, and devices using those systems, such as smart speakers and AI assistants, are now in wide use. However, speech recognition systems have security vulnerabilities. In particular, a known machine learning vulnerabi...

Full description

Saved in:
Bibliographic Details
Main Authors: Yuya Tarutani, Taisei Yamamoto, Yukinobu Fukushima, Tokumi Yokohira
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10693428/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850283090993217536
author Yuya Tarutani
Taisei Yamamoto
Yukinobu Fukushima
Tokumi Yokohira
author_facet Yuya Tarutani
Taisei Yamamoto
Yukinobu Fukushima
Tokumi Yokohira
author_sort Yuya Tarutani
collection DOAJ
description Machine learning technologies have improved the accuracy of speech recognition systems, and devices using those systems, such as smart speakers and AI assistants, are now in wide use. However, speech recognition systems have security vulnerabilities. In particular, a known machine learning vulnerability called audio adversarial examples (AAEs), which causes misrecognition in speech recognition systems, has become a problem. We propose a scheme for using speech processing to protect speech recognition systems from AAEs, preventing misrecognitions by slight processing of input speech that does not affect the recognition of normal speech. We use two kinds of processing: speed and frequency. Evaluation results show that the proposed scheme can reduce the success rate of attack speech to about 1% while maintaining about 85% recognition rates for normal speech.
format Article
id doaj-art-2c6db9fca6944cc3b7ab51179d484d91
institution OA Journals
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-2c6db9fca6944cc3b7ab51179d484d912025-08-20T01:47:50ZengIEEEIEEE Access2169-35362024-01-011214655114655910.1109/ACCESS.2024.346722410693428A Protection Scheme With Speech Processing Against Audio Adversarial ExamplesYuya Tarutani0https://orcid.org/0009-0006-9738-4750Taisei Yamamoto1Yukinobu Fukushima2Tokumi Yokohira3https://orcid.org/0009-0005-9891-9625Faculty of Interdisciplinary Science and Engineering in Health Systems, Okayama University, Okayama, JapanGraduate School of Interdisciplinary Science and Engineering in Health Systems, Okayama University, Okayama, JapanFaculty of Environmental, Life, Natural Science and Technology, Okayama University, Okayama, JapanFaculty of Interdisciplinary Science and Engineering in Health Systems, Okayama University, Okayama, JapanMachine learning technologies have improved the accuracy of speech recognition systems, and devices using those systems, such as smart speakers and AI assistants, are now in wide use. However, speech recognition systems have security vulnerabilities. In particular, a known machine learning vulnerability called audio adversarial examples (AAEs), which causes misrecognition in speech recognition systems, has become a problem. We propose a scheme for using speech processing to protect speech recognition systems from AAEs, preventing misrecognitions by slight processing of input speech that does not affect the recognition of normal speech. We use two kinds of processing: speed and frequency. Evaluation results show that the proposed scheme can reduce the success rate of attack speech to about 1% while maintaining about 85% recognition rates for normal speech.https://ieeexplore.ieee.org/document/10693428/Speech recognition systemsecurityaudio adversarial example
spellingShingle Yuya Tarutani
Taisei Yamamoto
Yukinobu Fukushima
Tokumi Yokohira
A Protection Scheme With Speech Processing Against Audio Adversarial Examples
IEEE Access
Speech recognition system
security
audio adversarial example
title A Protection Scheme With Speech Processing Against Audio Adversarial Examples
title_full A Protection Scheme With Speech Processing Against Audio Adversarial Examples
title_fullStr A Protection Scheme With Speech Processing Against Audio Adversarial Examples
title_full_unstemmed A Protection Scheme With Speech Processing Against Audio Adversarial Examples
title_short A Protection Scheme With Speech Processing Against Audio Adversarial Examples
title_sort protection scheme with speech processing against audio adversarial examples
topic Speech recognition system
security
audio adversarial example
url https://ieeexplore.ieee.org/document/10693428/
work_keys_str_mv AT yuyatarutani aprotectionschemewithspeechprocessingagainstaudioadversarialexamples
AT taiseiyamamoto aprotectionschemewithspeechprocessingagainstaudioadversarialexamples
AT yukinobufukushima aprotectionschemewithspeechprocessingagainstaudioadversarialexamples
AT tokumiyokohira aprotectionschemewithspeechprocessingagainstaudioadversarialexamples
AT yuyatarutani protectionschemewithspeechprocessingagainstaudioadversarialexamples
AT taiseiyamamoto protectionschemewithspeechprocessingagainstaudioadversarialexamples
AT yukinobufukushima protectionschemewithspeechprocessingagainstaudioadversarialexamples
AT tokumiyokohira protectionschemewithspeechprocessingagainstaudioadversarialexamples