NeurosciencebioRxivPreprint
Training neural networks to recognize speech increased their correspondence to the human auditory pathway but did not yield a shared hierarchy of acoustic features
Jessica A. F. Thompson, Yoshua Bengio, Elia Formisano, Marc Schönwiesner
This paper compares CNN speech recognition layers to human auditory fMRI activity and finds no shared acoustic feature hierarchy, with all regions most similar to the first fully-connected layer.