Bottleneck and Embedding Representation of Speech for DNN-based Language and Speaker Recognition | Publicación