A Fly on the Wall -- Exploiting Acoustic Side-Channels in Differential Pressure Sensors

TL;DR

BaroVox reveals that differential pressure sensors, when placed near acoustic sources, inadvertently capture speech via diaphragm vibrations and can be transformed into an eavesdropping channel. By modeling the pressure-sound interaction with a Pressure-Acoustic Transformation (PAT) and deploying two attack designs—DS-I (signal-processing) and DS-II (DL-based ASR)—the work demonstrates both manual and automatic recovery of speech from DPS outputs, achieving a mean word error rate of $0.29$ and an automation accuracy of $90.51\%$. The study validates feasibility in cleanroom-relevant settings, highlights privacy and IP implications across industries, and proposes practical countermeasures such as sound dampening, filtering, and distance management. The findings underscore the need for secure data handling and defense strategies as DPS become increasingly integrated into IoT and smart-building ecosystems.

Abstract

Differential Pressure Sensors are widely deployed to monitor critical environments. However, our research unveils a previously overlooked vulnerability: their high sensitivity to pressure variations makes them susceptible to acoustic side-channel attacks. We demonstrate that the pressure-sensing diaphragms in DPS can inadvertently capture subtle air vibrations caused by speech, which propagate through the sensor's components and affect the pressure readings. Exploiting this discovery, we introduce BaroVox, a novel attack that reconstructs speech from DPS readings, effectively turning DPS into a "fly on the wall." We model the effect of sound on DPS, exploring the limits and challenges of acoustic leakage. To overcome these challenges, we propose two solutions: a signal-processing approach using a unique spectral subtraction method and a deep learning-based approach for keyword classification. Evaluations under various conditions demonstrate BaroVox's effectiveness, achieving a word error rate of 0.29 for manual recognition and 90.51% accuracy for automatic recognition. Our findings highlight the significant privacy implications of this vulnerability. We also discuss potential defense strategies to mitigate the risks posed by BaroVox.

Figures (12)

• Figure 1: Components of a DPS.
• Figure 2: Components of a semiconductor cleanroom.
• Figure 3: Overview of the attack model.
• Figure 4: Our experimental setup for a feasibility study.
• Figure 5: STFT plot of a spoken word "one".