Acquisition of Audio Information from Silent High Speed Video

A method to construct audio information from silent high speed video is presented. While an audio sample is played, an object’s small vibrations generated by the sound waves are recorded by a single high-speed camera at high frame rates. Images are then analyzed using a Digital Image Correlation (DIC) scheme, where subsets of pixels are observed for displacement in two dimensions. The displacement data can be mapped to audio signals over time. Prior DIC analysis applications rely on artificial speckle patterns for observations, whereas the proposed method can employ natural patterns on objects, such as those on a piece of newspaper or a myriad of other common objects. Thin, flexible, and lightweight objects were found to output audio signals with less audible noise than other objects. Experiments have concluded that this method is a rapid and effective dynamic measurement process that results in acquisition of audio information that was not captured during high-speed recording.