Chapter 9: Software 2.0 | Deep Learning: Teaching Computers to Hear

9.1 Sound Is Just a Picture

In Chapter 8 you taught a computer to see. Now you'll teach it to hear, and the trick is smaller than you'd expect: you turn the sound into a picture first. A guitar note becomes an image called a spectrogram, and once it's a picture, the same kind of model that recognized objects can recognize chords. Hearing, for a computer, is mostly seeing.