Predictive speech enhancement models can be manipulated by psychoacoustically masked adversarial noise to alter output semantics, while diffusion models exhibit inherent robustness.
A well-known example is the addition of adversarial noise to an image of a panda so that a classifier would detect a gibbon instead [1]
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Are Modern Speech Enhancement Systems Vulnerable to Adversarial Attacks?
Predictive speech enhancement models can be manipulated by psychoacoustically masked adversarial noise to alter output semantics, while diffusion models exhibit inherent robustness.