Computer scientists at MIT have opened new doors in speech and image recognition systems by creating a model system that can identify objects within an image based solely on the spoken descriptions of ...