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

A targeted adversarial attack produces audio samples that can force anAutomatic Speech Recognition (ASR) system to output attacker-chosen text. Toexploit ASR models in real-world, black-box settings, an adversary can leveragethe transferability property, i.e. that an adversarial sample produced for aproxy ASR can also fool a different remote ASR. However recent work has shownthat transferability against large ASR models is very difficult. In this work,we show that modern ASR architectures, specifically ones based onSelf-Supervised Learning, are in fact vulnerable to transferability. Wesuccessfully demonstrate this phenomenon by evaluating state-of-the-artself-supervised ASR models like Wav2Vec2, HuBERT, Data2Vec and WavLM. We showthat with low-level additive noise achieving a 30dB Signal-Noise Ratio, we canachieve target transferability with up to 80% accuracy. Next, we 1) use anablation study to show that Self-Supervised learning is the main cause of thatphenomenon, and 2) we provide an explanation for this phenomenon. Through thiswe show that modern ASR architectures are uniquely vulnerable to adversarialsecurity threats.

Watch What You Pretrain For: Targeted, Transferable Adversarial Examples on Self-Supervised Speech Recognition models