# Universal Adversarial Perturbations for Speech Recognition Systems

## TL;DR

This paper introduces universal adversarial audio perturbations that can fool speech recognition systems across different models, highlighting vulnerabilities in current ASR technology.

## Contribution

We propose a novel algorithm to generate quasi-imperceptible universal audio perturbations effective against multiple ASR systems.

## Key findings

- Universal perturbations successfully fool state-of-the-art ASR systems.
- Perturbations transfer across different models, indicating broad vulnerability.
- Audio-agnostic perturbations can be crafted with minimal perceptual impact.

## Abstract

In this work, we demonstrate the existence of universal adversarial audio perturbations that cause mis-transcription of audio signals by automatic speech recognition (ASR) systems. We propose an algorithm to find a single quasi-imperceptible perturbation, which when added to any arbitrary speech signal, will most likely fool the victim speech recognition model. Our experiments demonstrate the application of our proposed technique by crafting audio-agnostic universal perturbations for the state-of-the-art ASR system -- Mozilla DeepSpeech. Additionally, we show that such perturbations generalize to a significant extent across models that are not available during training, by performing a transferability test on a WaveNet based ASR system.