English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Automatic Speech Recognition services (ASRs) inherit deep neural networks'vulnerabilities like crafted adversarial examples. Existing methods oftensuffer from low efficiency because the target phases are added to the entireaudio sample, resulting in high demand for computational resources. This paperproposes a novel scheme named FAAG as an iterative optimization-based method togenerate targeted adversarial examples quickly. By injecting the noise over thebeginning part of the audio, FAAG generates adversarial audio in high qualitywith a high success rate timely. Specifically, we use audio's logits output tomap each character in the transcription to an approximate position of theaudio's frame. Thus, an adversarial example can be generated by FAAG inapproximately two minutes using CPUs only and around ten seconds with one GPUwhile maintaining an average success rate over 85%. Specifically, the FAAGmethod can speed up around 60% compared with the baseline method during theadversarial example generation process. Furthermore, we found that appendingbenign audio to any suspicious examples can effectively defend against thetargeted adversarial attack. We hope that this work paves the way for inventingnew adversarial attacks against speech recognition with computationalconstraints.

FAAG: Fast Adversarial Audio Generation through Interactive Attack Optimisation