Recognizing emotion from singing and speaking using shared models | IEEE Conference Publication | IEEE Xplore