Acoustic feature prediction from semantic features for expressive speech using deep neural networks | IEEE Conference Publication | IEEE Xplore