Precision-Varying Prediction (PVP): Robustifying ASR systems against adversarial attacks
This work addresses adversarial vulnerabilities in ASR systems, which is critical for deployed automated systems, but it is incremental as it builds on existing precision-based insights.
The paper tackles the problem of adversarial robustness in automatic speech recognition (ASR) systems by introducing Precision-Varying Prediction (PVP), which randomly samples model precision during inference to reduce attack success; experiments show a significant increase in robustness and competitive detection performance across various models and attacks.
With the increasing deployment of automated and agentic systems, ensuring the adversarial robustness of automatic speech recognition (ASR) models has become critical. We observe that changing the precision of an ASR model during inference reduces the likelihood of adversarial attacks succeeding. We take advantage of this fact to make the models more robust by simple random sampling of the precision during prediction. Moreover, the insight can be turned into an adversarial example detection strategy by comparing outputs resulting from different precisions and leveraging a simple Gaussian classifier. An experimental analysis demonstrates a significant increase in robustness and competitive detection performance for various ASR models and attack types.