This study presents a new system designed to protect against eavesdropping by jamming microphones with a unique phoneme-based noise. This noise is designed to distract both humans and machines and is resistant to denoising techniques. The system also allows for authorized recording, expanding its potential uses. Experimental results show that this system can reduce the accuracy of speech recognition systems to below 50%. This represents an improvement over existing anti-eavesdropping solutions.
Publication date: 31 Jan 2024
Project Page: N/A
Paper: https://arxiv.org/pdf/2401.15704
