August 29, 2024




Track 1

Silencing Spies: Revealing Microphone Nonlinear Behavior and Building an Ultrasonic Jammer


Nisl Lab, Tsinghua University

Senior Engineer

Undisclosed Co

In the contemporary digital age, where privacy and information security are under constant threat, the ubiquitous presence of sound sensors in devices has become a double-edged sword. These sensors, while integral to modern conveniences, also pose a significant risk as malicious entities can exploit them to monitor and record private activities covertly. This emerging threat landscape demands innovative countermeasures to safeguard individual and organizational privacy.

Our research introduces a novelty ultrasonic jammer, representing a significant advancement in counter-eavesdropping technology. This device transcends traditional reactive measures by combining a sophisticated Dynamic Intermodulation Modulation (DIM) algorithm with an innovative design of an ultrasonic speaker array. The DIM algorithm, a refined evolution of our initial model, exploits the nonlinear phenomena in microphones under high sound pressure, effectively neutralizing unauthorized listening across a variety of devices such as Electret Condenser Microphones (ECM), Micro-Electro-Mechanical Systems (MEMS), and piezoelectric ceramics by amplifying their nonlinear frequency response.

Complementing this, the jammer’s ultrasonic speaker array, composed of multiple high-frequency transducers strategically positioned, generates a dense, impenetrable field of ultrasonic waves, ensuring widespread disruption of spying devices. Engineered for adaptability in diverse environmental conditions, our solution’s modular speaker design allows for scalable and customized deployment. By employing advanced materials and manufacturing techniques, we’ve enhanced the durability and performance of the array, guaranteeing long-term efficacy in privacy protection and providing a comprehensive solution against sophisticated surveillance tactics.

In our testing phase, we assessed the robustness of our ultrasonic jammer under diverse scenarios. This included different placements of eavesdropping devices, comparisons with existing commercial devices, and conditions with potential physical obstructions. Our experiments showed a significant enhancement in performance compared to alternatives in the market, with a disruption rate nearing 100% under similar conditions.

Furthermore, when subjected to advanced Speech-to-Text (STT) models like iFLYTEK and, the jammer successfully rendered speech indecipherable, affirming its effectiveness in thwarting eavesdropping attempts. Additionally, considering the need for a more compact and portable jammer during interactions between two people or in small groups, we reduced the size of the array and decreased the testing distance.

Under these conditions, the interference remained highly effective, making the speech undetectable by both humans and AI STT models.