Internal noise from a CCD in a light tight box is mapped to audio by sampling pixels in a Quicktime matrix and using those values to manipulate white noise. Software looks for similarity between spectrograms of the noise and a very large set of spectrograms of spoken words. When two are congruent enough the results are spoken by the computer and projected into the space. Software pans across, zooms, and changes the blur, brightness, and contrast of the camera noise, as it looks for hidden images. These are standard EVP strategies for teasing images and sounds from visual and audio noise.” – alan dunning and paul woodrow

« »