SER

A task and associated evaluation methodology in affective computing that focuses on automatically identifying a speaker's emotional state from speech signals. It is a core evaluation direction in the field of affective computing.