Computer scientists at MIT have developed a machine-learning system that can identify objects in an image based on a spoken description of the image. Typical speech recognition systems like Google ...