New Technique Permits Researchers To Extract Audio From Silent Video

New Technique Allows Researchers To Extract Audio From Silent Video

By wanting on the minute vibrations in extreme velocity video, researchers on the Division of Engineering of the Catholic School of America have created a method for extracting sound information from extreme-velocity silent video footage.

From the launch:

The authors used a subset-based image-correlation technique to detect the motions of things on the ground of an object, capturing aim footage with a extreme-velocity digital digital camera and making use of the Gauss-Newton algorithm and a few totally different measures to understand very fast and very right image matching. Because of the detected vibrations are immediately related to sound waves, a straightforward model was used to reconstruct the distinctive audio information of the sound waves.

Whereas this wouldn’t work for, say, extracting sound from earlier silent film, the researchers say they could extract conversations from vibrations at a “far distance,” a prospect that is instantly fascinating and a bit scary.

The paper, “Audio extraction from silent extreme-velocity video using an optical technique,” appeared inside the journal Optical Engineering.

This technique is simply not new – MIT researchers have been able to study sound from a stray potato chip bag – nevertheless this system makes use of a neater image matching system to sense audio.