InjectEave Attack Recovers Audio from Headphones Up to 30 Meters Away

Key Takeaways

• A new electromagnetic (EM) attack, dubbed InjectEave, allows adversaries to remotely eavesdrop on audio playing through both wired and wireless headphones.
• The attack leverages an active radio-frequency (RF) injection technique, enabling surveillance from up to 30 meters away, even through concrete walls, using commercially available equipment.
• Unlike traditional passive EM side-channel attacks, InjectEave actively induces leakage by interacting with nonlinear hardware components common in consumer electronics.
• Beyond audio eavesdropping, the technique can infer usage patterns of smart home devices and has been demonstrated in a real-time, AI-powered voice spoofing scenario against a landline phone.
• Current EM shielding and cryptographic defenses are largely ineffective, necessitating a re-evaluation of hardware-software co-design for analog security.

InjectEave: A New Threat to Audio Privacy via Electromagnetic Injection

Cybersecurity researchers have unveiled a groundbreaking electromagnetic (EM) attack, named InjectEave, that enables sophisticated eavesdropping on audio from wired and wireless headphones. This novel technique allows an attacker to recover audio playing through headphones from distances of up to 30 meters, even penetrating physical barriers like walls, utilizing readily available radio-frequency (RF) equipment.

The research, conducted by a collaborative team from the Hong Kong University of Science and Technology (Guangzhou) and the Hong Kong Polytechnic University, details the method in their paper, “Injected and Leaked: Actively Inducing Side-Channel Leakage Using Electromagnetic Injection and Hardware Nonlinearity,” which has been accepted at USENIX Security ’26.

How InjectEave Operates: Active Signal Manipulation

InjectEave fundamentally differs from conventional EM side-channel attacks, which passively monitor incidental electromagnetic emissions. Instead, InjectEave actively injects a precisely tuned radio-frequency carrier signal into the target device. This injected signal then interacts with inherent nonlinear hardware components present in common electronics, such as audio amplifiers, analog-to-digital converters (ADCs), power converters, and switching MOSFETs. These components are ubiquitous in headphones, smart home devices, and traditional landline phones.

The interaction causes these components to inadvertently “mix” the confidential analog signal, such as an audio waveform, with the injected carrier. This process effectively modulates the low-frequency secret onto a much higher-frequency EM wave. This modulated signal can then radiate significantly further than what would be possible with passive eavesdropping methods, overcoming previous physical limitations.

The researchers characterize this mechanism as an “Injection-Modulation-Emission” model. In this model, the injected carrier couples into the device’s internal circuitry, blends with the secret signal through hardware nonlinearity, and subsequently re-radiates from the same wires and cables that initially served as unintentional antennas. A critical finding is that the strength of the leaked signal scales almost linearly with both the amplitude of the secret signal and the attacker’s injection power, implying that increased transmission power directly enhances the strength and recoverability of the eavesdropped signal.

Overcoming Physical Barriers and Distance Limitations

Traditional EM side-channel eavesdropping has historically faced a significant physical constraint: the vast frequency mismatch between human speech (20 Hz to 20 kHz) and the megahertz-to-gigahertz range required for efficient EM radiation from device wiring. This discrepancy limited passive audio eavesdropping distances to less than 1.5 meters in prior studies like MagEar and Periscope. InjectEave bypasses this fundamental barrier by actively shaping the leakage rather than merely detecting it, allowing attackers to optimize the injection frequency for specific target devices.

The research team employed an Ettus USRP B210 software-defined radio, log-periodic antennas, and a spectrum analyzer to test InjectEave against 11 commercial off-the-shelf devices. The tested devices included wired headphones from Sony, Dell, and Apple, wireless headphones from UGreen, Philips, and HP, a Flyingvoice VoIP landline phone, and smart fans and lamps from Xiaomi and other manufacturers.

At a baseline distance of 50 centimeters, the attack achieved nearly 100% audio recognition rates across almost all audio devices. Signal-to-noise ratios (SNR) ranged from approximately 6 dB for Apple Earbuds to over 23 dB for UGreen MAX2 wireless headphones.

By incorporating an external power amplifier, costing approximately $415, the injection power was boosted from 18 dBm to 40 dBm. This enhancement dramatically extended the effective eavesdropping range for the UGreen MAX2 and Philips TAH2020 headphones to an impressive 30 meters, while maintaining intelligible speech recovery. Furthermore, the attack demonstrated robustness against physical obstructions: glass and wood caused negligible signal loss (1 to 2 dB), and even solid concrete walls only attenuated the signal by about 5.8 dB for headphones, enabling reliable through-wall eavesdropping in various environments such as offices, hotels, and conference rooms.

The technical disclosure published in ArXiv details how a teardown analysis by the researchers revealed that twisted-pair cabling could reduce leakage by up to 20 dB compared to standard parallel wiring, offering a potential, albeit partial, mitigation strategy for hardware manufacturers.

Advanced Applications: Voice Spoofing and IoT Surveillance

Perhaps one of the most concerning demonstrations involved a closed-loop “Eavesdrop-Synthesize-Inject” attack against a Flyingvoice landline phone. In this scenario, an attacker first eavesdrops on one party’s voice to gather contextual information. Upon detecting trigger keywords, such as “quote” or “confirmation,” an AI voice-cloning tool like IndexTTS-2 is used to synthesize a deepfaked version of the speaker’s voice. This synthesized audio is then injected back into the target’s headset in real time. Testing indicated that the injected fake audio was nearly indistinguishable from the original speaker’s voice, exhibiting only a 0.071 average deviation in a standard speech-intelligibility metric.

To enhance the clarity of the recovered audio, which inherently suffers from noise and harmonic distortion, the InjectEave signal-enhancement module incorporates a diffusion-based speech-denoising model. This model was trained on physics-simulated data, successfully boosting the recovered audio’s signal-to-noise ratio from 7.0 dB to 16.1 dB and improving its intelligibility score from 0.58 to 0.72.

The scope of InjectEave extends beyond headphones. The same principle was used to expose power consumption patterns of smart lamps and fans, allowing attackers to remotely infer household activity patterns, sleep schedules, and appliance usage without requiring any network access. The researchers also found preliminary evidence that microphone inputs are vulnerable, though current limitations restrict the effective range to approximately 30 centimeters due to microphones’ significantly weaker signal levels.

Current defensive measures, such as cryptographic masking and EM shielding, prove largely ineffective against InjectEave. This is because the attack targets continuous analog signals rather than digital logic, and a determined attacker can simply increase injection power to overcome standard shielding. The researchers emphasize that a comprehensive, security-conscious hardware-software co-design approach will be essential to address this newly exposed analog attack surface across headphones, IoT devices, and other consumer electronics.

What You Should Do

• Be Aware of Your Surroundings: Understand that devices with analog components, like headphones and smart home gadgets, can be vulnerable to remote eavesdropping even through walls.
• Consider Physical Distance: While difficult to completely mitigate, maintaining greater physical distance from potential attackers may reduce the likelihood of successful audio recovery, especially for less powerful adversaries.
• Advocate for Secure Hardware Design: Support manufacturers who prioritize security-aware hardware-software co-design. For hardware developers, consider implementing twisted-pair cabling where feasible, as it demonstrated reduced leakage.
• Exercise Caution with Sensitive Conversations: For highly confidential discussions, consider environments free of electronic devices or employ white noise generators, although their effectiveness against InjectEave is not fully established.
• Stay Informed: Keep abreast of security research and product updates from device manufacturers regarding new vulnerabilities and potential mitigations.

Disclaimer: HackersRadar reports on cybersecurity threats and incidents for informational and awareness purposes only. We do not engage in hacking activities, data exfiltration, or the hosting or distribution of stolen or leaked information. All content is based on publicly available sources.