Integrating linguistic cues into speech-based emotion recognition

The integration of emotions in the concept of spoken language dialog systems has attracted a lot of interest in current research on human-computer interfaces during the past years. As a supplement to the vast variety of existing emotion recognition systems using different modalities, we propose a straightforward approach to derive a speaker's emotional state from the linguistic content of respective utterances. In this article, we describe the use of what we refer to as "affective grammars" and we also pick up and advance the idea of using Hidden Markov Models to jointly recognize speech and emotions from the same speech signals. Then we outline an integrative approach how these linguistic cues can be used to enhance the robustness of the emotion recognizer with particular regards to their use in the dialog success estimation in call centers.