Have a group individually describe a series of simple images under a time constraint. Set a threshold for frequency changes in each description to determine 'mo

Have a group individually describe a series of simple images under a time constraint. Set a threshold for frequency changes in each description to determine 'more musical' or 'less musical' entries (voice to MIDI?). Shuffle the descriptions, have a second group individually listen to one description for each image and draw a picture under time constraint. Shuffle the drawings, display 2 at a time and the original image and have a third group individually vote for which comes closer to the source image. I've never designed an academic study just contributing - I'd assume drawing ability is a big weak point. Or maybe show the original image in addition to similar images (with one key element different in each one) and have the second group select the image that most closely resembles the description.