METHOD AND DEVICE FOR DETECTING EMOTION

PROBLEM TO BE SOLVED: To increase the precision of emotion detection processing and to reduce the possibility of an error occurrence. SOLUTION: According to a specific input voice, at least some of features of at least 1st and 2nd feature classes are evaluated, derived, and/or extracted and the current emotional state of a speaker and/or parameters of the emotional state are derived from the features. The 1st and 2nd feature classes are allocated to individual dimensions of an underlying emotional manifold or emotional space, such as an awakening degree and comfortableness, and/or with subspaces thereof. COPYRIGHT: (C)2003,JPO