In the largest study yet of deep neural networks trained to perform auditory tasks, researchers found most of these models generate internal representations that share properties of representations seen in the human brain when people are listening to the same sounds.
