Hierarchical spectro-temporal features for robust speech recognitionXavier DomontAgotado4,3Notifícame