Acoustic Attacks: How Speakers Can Covertly Hack Computers Without Physical Contact

An Attack That Requires No Physical Contact

In the traditional cybersecurity mindset, attackers either penetrate remotely through networks or need physical access to the target device—inserting USB drives, connecting cables, or directly operating the keyboard. A concept project called Pwnd Blaster proposes a third path: achieving intrusion entirely through sound waves emitted by speakers, without touching any hardware interface on the computer.

This technical share from the Reddit community reveals an easily overlooked attack surface—the acoustic channel between devices. Attackers don't need to connect to WiFi, don't need Bluetooth pairing, and don't even need the target device to actively run any networked service. Sound alone can potentially establish a covert data channel.

How Sound Waves Become an Attack Vector

Basic Principles of Covert Acoustic Channels

The core of acoustic attacks lies in using sound waves as a medium for information transmission. A computer's speakers and microphones are essentially transducers that convert between electrical signals and sound waves. The fundamental working principle of a transducer is converting one form of energy into another: speakers use electromagnetic force to drive a diaphragm, converting electrical signals into air vibrations (sound waves), while microphones perform the reverse operation, capturing sound waves and converting them back into electrical signals. This bidirectional conversion capability means that any device equipped with both components inherently possesses the hardware foundation for acoustic communication.

Attackers can encode data into sound signals at specific frequencies—including ultrasonic frequency bands inaudible to the human ear (typically above 20kHz)—transmit them through one device's speaker, and have them received and decoded by the target device's microphone. It's worth noting that while the upper limit of human hearing is typically around 20kHz, most commercial speakers and microphones have actual frequency response ranges that extend to 22kHz or even above 24kHz. This means there exists a "near-ultrasonic" window (18kHz-24kHz) where signals are virtually imperceptible to most people but still within the operating range of consumer-grade audio hardware. For data encoding, attackers can employ modulation techniques such as Frequency-Shift Keying (FSK) and Orthogonal Frequency-Division Multiplexing (OFDM) to map binary data onto different frequencies or phases, enabling reliable acoustic data transmission.

This type of technology is not new in academia. Previous notable research such as BadBIOS (though highly controversial) and a series of "air-gap attacks" studies from Ben-Gurion University of the Negev in Israel have all explored transmitting data across network isolation through physical channels including sound, light, electromagnetic radiation, and heat. Dr. Mordechai Guri and his team at Ben-Gurion University have published over 30 different air-gap bridging techniques since 2014, with acoustic-class attacks including "DiskFiltration" (using hard drive actuator arm vibrations to generate sound), "Fansmitter" (using fan speed variations to convey information), and "MOSQUITO" (directly using speaker ultrasound for transmission). These studies prove that virtually any computer component capable of producing or sensing physical signals can potentially be weaponized as a covert communication channel. Pwnd Blaster can be seen as an extension of this research lineage toward practical application.

Why "No Contact" Defeats Defenses

For scenarios with extremely high security requirements—such as "air-gapped" systems physically isolated from the internet—traditional protective measures primarily focus on network boundaries and physical interfaces. An "air gap" refers to the security strategy of completely physically disconnecting sensitive computing systems from any external network, widely used in military command and control systems, nuclear facility industrial control networks, financial core clearing systems, and classified government agencies. The security model of these systems is built on a core assumption: as long as all network connections and wireless communication capabilities are severed, external attackers cannot remotely interact with the system. Traditional air-gap protections typically include disabling wireless network cards, blocking USB ports, physically removing Bluetooth modules, and even implementing electromagnetic shielding (Faraday cages) around server rooms.

But if an attack can be accomplished purely through sound waves, then even the most rigorous network firewalls and strictest USB port controls may prove ineffective. Sound waves, as mechanical waves, propagate freely through air, unblocked by electromagnetic shielding (Faraday cages can block electromagnetic signals but cannot prevent sound propagation), and leave no network logs or packet records. This means traditional security monitoring systems—including Intrusion Detection Systems (IDS), Security Information and Event Management platforms (SIEM), and network traffic analysis tools—are completely "blind" to such attacks.

As long as the target device is equipped with a microphone (nearly all laptops today have built-in microphones) and has been pre-compromised with malware capable of monitoring acoustic channels, attackers can covertly issue commands or exfiltrate data within a certain distance, leaving no cable traces and remaining virtually undetectable by conventional traffic monitoring.

Real-World Constraints of the Acoustic Attack Chain

It's important to take a rational view: while such acoustic attacks are conceptually striking, they face significant limitations in practical implementation.

Prerequisites Remain Demanding

First, most acoustic channel attacks are not "attacks from thin air." The target device typically needs to be pre-compromised with a malicious program capable of receiving and decoding acoustic commands. Sound waves primarily serve as a covert control channel or data exfiltration channel for already-infected devices, rather than an initial intrusion method from scratch. In cybersecurity terminology, this type of remote control channel is called a C2 (Command and Control) channel—attackers use it to send commands to implanted malware and retrieve stolen data. Traditional C2 channels typically rely on network pathways such as HTTP/HTTPS, DNS tunneling, or social media platforms, while the unique value of an acoustic C2 is its ability to operate in completely disconnected environments.

In other words, sound waves solve the problem of "how to communicate in an isolated environment," not "how to breach a system for the first time." So how is the initial malware implanted into air-gapped systems? Historical cases show common vectors include: supply chain attacks (implanting backdoors before equipment leaves the factory or during transit), insider threats (intentionally or unintentionally bringing infected media into isolated zones), and leveraging maintenance windows to spread via removable storage media (the famous Stuxnet worm crossed the air gap of Iran's nuclear facilities via USB flash drives). Once malware successfully takes residence, the acoustic channel becomes the attacker's "last mile" solution for maintaining communication.

Constraints of Distance, Bandwidth, and Environmental Noise

Second, acoustic channel transmission rates are extremely low—often only a few bits to tens of bits per second—and are highly sensitive to distance and environmental noise. For comparison, even early 56K dial-up modems could provide 56,000 bits per second of bandwidth, and modern WiFi connections operate at hundreds of megabits or even gigabits per second. Acoustic channel bandwidth is 3-4 orders of magnitude lower than even the slowest network connection, fundamentally limiting its application scenarios.

From a physics perspective, ultrasound attenuation in air is far more severe than audible sound. The atmospheric absorption coefficient of sound waves is approximately proportional to the square of the frequency. Ultrasound at 20kHz experiences attenuation of about 0.5-1dB per meter in room-temperature dry air, meaning signal strength drops significantly over just a few meters. Additionally, office equipment (such as monitors, switching power supplies, and fluorescent lights) generates environmental noise in the ultrasonic frequency band, further compressing the available signal-to-noise ratio.

In noisy office environments, ultrasonic signals can easily be weakened or interfered with. Therefore, this type of attack is better suited for stealing small but high-value data like passwords and cryptographic keys, rather than large-scale file transfers. Take a 256-bit encryption key as an example: even at an extremely low rate of 10 bits per second, exfiltration would take only about 26 seconds—already enough to cause catastrophic security consequences.

Security Recommendations Against Acoustic Attacks

Despite the many limitations, projects like Pwnd Blaster still carry important cautionary significance. They remind security practitioners that the attack surface is far broader than we imagine.

For high-security environments, the defensive mindset should extend to the physical layer:

• Disable or physically remove unnecessary microphones and speakers, especially on air-gapped systems;
• Deploy ultrasonic detection equipment to monitor acoustic activity in anomalous frequency bands;
• Implement electromagnetic and acoustic shielding for sensitive devices—it's important to emphasize here that acoustic shielding and electromagnetic shielding require different technical approaches: electromagnetic shielding uses conductive materials in a Faraday cage configuration, while acoustic shielding requires soundproofing materials, anechoic structures, or active noise cancellation technology. Both must be deployed in combination to achieve comprehensive protection;
• Strengthen endpoint detection to promptly discover malicious processes that may act as acoustic channel receivers—specifically, security teams should monitor abnormal microphone access behavior, detect suspicious programs performing audio processing in ultrasonic frequency bands, and audit anomalous API calls at the audio driver layer.

For ordinary users, the threshold and prerequisites for such attacks are high, so there's no need for excessive alarm. However, it does illustrate an important principle: any sensor capable of receiving external signals can potentially become an intrusion entry point.

Conclusion

The acoustic intrusion technique demonstrated by Pwnd Blaster pulls the cybersecurity discussion back from the software and network layers to the physical world. It's not meant to create panic, but rather to reveal a long-underestimated attack dimension. In an era of ubiquitous connectivity and omnipresent sensors, the boundaries of security defense need to continuously extend outward—because attackers are always searching for those overlooked gaps.