Is there no industry safe from economic encroachment by automation and machine learning? A team from the University of Toronto have built a digital Irving Berlin that can generate Christmas carols from a single image.
The Toronto researchers relied on a pair of neural networks to create the AI. The first network was trained in the art of carolling with a hundred hours of online music. This enabled it to generate a basic 120 BPM melody -- complete with chords and drums -- based on a musical scale and melodic profile. To write the lyrics, a second neural net was shown a picture of a Christmas tree, which served as the song's subject. Put those elements together and you've got yourself a cheerful holiday ditty perfect for listening to while waiting out Robot Santa Claus' annual reign of terror.