All this is the product of a relatively simple AI mechanism. It uses object recognition to identify elements within the image, and then matches these to a dataset of more than 52,000 sound files ...