Exploiting prosodic and lexical features for tone modeling in a conditional random field framework | IEEE Conference Publication | IEEE Xplore