search menu icon-carat-right cmu-wordmark

Emotion Recognition from Voice in the Wild

This poster represents research to determine the feasibility of using micro-articulometry technique to accurately recognize emotion from voice.

Software Engineering Institute


Accurately recognizing emotion from voice is important in defense applications such as speaker profiling and human-machine teaming, but is currently infeasible. We introduce a new, continuous speech emotion recognition database, CMU-SER, and a set of micro-articulometry techniques that can capture finer nuances than the current state of the art.