This video was made a group of security researchers based at the Ben-Gurion University of the Negev and the Weizmann Institute of Science in Israel. The Lamphone, as they call it, is intended as an alternative method of eavesdropping on private conversations without having to compromise a device with malware. In their tests, the researchers were able to accurately monitor audio, including speech and music, from about 80 feet away; they think they could amplify that range with some better hardware, too.
And all it takes is a few simple tools:
- Telescope – This piece of equipment is used to focus the field of view on the hanging bulb from a distance.
- Electro-optical sensor – This sensor is mounted on the telescope and consists of a photodiode (a semiconductor device) that converts light into an electrical current. The current is generated when photons are absorbed in the photodiode. Photodiodes are used in many consumer electronic devices (e.g., smoke detectors, medical devices).
- Sound recovery system – This system receives an optical signal as input and outputs the recovered acoustic signal. The eavesdropper can implement such a system with dedicated hardware (e.g., using capacitors, resistors, etc.). Alternatively, the attacker can use an ADC to sample the electro-optical sensor and process the data using a sound recovery algorithm running on a laptop.
So basically, a laser; something to point it at (like a light bulb); and something to convert the sound. It works with LEDs as well as incandescent bulbs, too.
Here's how Davey Winder at Forbes describes the process:
Fluctuations in air pressure on the surface of the hanging bulb are created by the sound of conversation, or music, and make a hanging bulb vibrate. An analog-to-digital converter makes the conversion from electrical to digital information. The researchers have developed an algorithm for the attack method, which they named Lamphone, that can "recover sound from the optical measurements obtained from the vibrations of a light bulb." This is done passively and without needing to be in the same room. When it comes to human speech, the researchers used Google Cloud Speech to transcribe the conversation, and the music was correctly identified using Shazam and SoundHound.
There are some quirks, however; you need a clear line of sight to the bulb, which means no curtains, blinds, or even lampshades. And the quality of laser-through-lightbulb-glass-vibration-audio-conversion will depend on the thickness of the glass bulb, and how close the conversation happens to it.
It's a pretty interesting technique. Also terrifying, but cool. Read the full paper below:
Lamphone: Real-Time Passive Sound Recovery from Light Bulb Vibrations [Ben Nassi, Yaron Pirutin, Adi Shamir, Yuval Elovici, Boris Zadov / Lamphone]
How Hackers Use An Ordinary Light Bulb To Spy On Conversations 80 Feet Away [Davey Winder / Forbes]