Salient Features for Anger Recognition in German and English IVR Portals

摘要: Anger recognition in speech dialogue systems can help to enhance human commputer interaction. In this chapter we report on the setup and performance opti-izationtechniques for successful anger classification using acoustic cues. We evaluate of a broad variety features both German an American English voice portal database which contain “real” (i.e. non-acted) continuous narrow-band quality. Starting with large-scale feature extraction, determine optimal sets combinations each language, by applying Information-Gain based ranking scheme. Analyzing notice that large proportion most promising databases are derived from MFCC loudness. contrast similarity also pitch proved importance database. further calculate scores our setups discriminative training Support-Vector Machine classification. The developed show