The International Society for Optics and Photonics reported recently that a new technique has been developed that allows for the extraction of audio information from silent high-speed video. It has been referred to as the “silent walls can now talk” technology.
The technology utilizes an image-matching procedure that hones in on the vibrations released from the high speed video sound waves. One of the researchers involved in the technology was quoted as saying, “One of the intriguing aspects of the paper is the ability to recover spoken words from a video of objects in the room,” said journal Associate Editor Reiner Eschbach, a Research Fellow at Xerox Corp. It showed that sound creates vibrations in objects and with the correct equipment, those vibrations can be detected through the use of a video signal.
Because sound waves cause the air around them to vibrate and that motion can cause objects in its path to pick up those vibrations (especially if the object is thin and flexible like a piece of paper) they can be detected and analyzed. Once the information is captured, it is possible to reconstruct the audio based on the captured algorithms.
Researchers set up the experiment in such a way that the sound waves were detected by high speed cameras using custom lenses on specific points on the surface of an object. The vibrations that are detected are those related to the sound waves and from that the researchers created a model to recreate the original audio sound waves.
Light travels further than sound through air and can also pass through glass and because of this the researchers believe they will eventually be able to have passive detection of conversations. The sensitivity and accuracy of the technology is still being refined and ways to remove potential interference from the testing metrics are also being designed.
UKA Optics is a manufacturer of standard and custom CCTV lens assemblies for image capture, barcode scanning, night vision systems, medical systems, hi-speed imaging, machine vision and robotics vision.