Humanoid robot and Artificial Intelligence make a proper mutual communication to interact with ordinary people. Advanced research in emotional interaction field will be based on the growing understanding, speech recognition and facial expression processes. Software specified for human-robot interaction needs to receive human emotions and optimize their behavior. In this paper, we report the results obtained from an exploratory study on software which automatically recognizes and classifies basic emotional states (sadness, surprise, happiness, anger, fear and disgust). The study consists of generating and analyzing the graphs of speech signals with using: Pitch, intensity and formant properties of emotive speech. Indeed, facial feature extraction phase uses the mathematical formulation and measures a set of Action Units (AUs) for emotion classification. The efficiency of the methodology was evaluated by experimental tests on 300 individuals (150 females and 150 males, 20 to 48 years old) multi-ethnic groups, namely: (i) European, (ii) Asian Middle East and (iii) American. In the light of the experiments, the accuracy of the proposed model for emotional detection time was calculated 2.53 s as we defined primarily more distinct boundaries between emotions to classify features in a set of basic emotions.
System and method for recognizing human emotion state based on analysis of speech and facial feature extraction; Applications to human-robot interaction
Gasparetto, Alessandro
2016-01-01
Abstract
Humanoid robot and Artificial Intelligence make a proper mutual communication to interact with ordinary people. Advanced research in emotional interaction field will be based on the growing understanding, speech recognition and facial expression processes. Software specified for human-robot interaction needs to receive human emotions and optimize their behavior. In this paper, we report the results obtained from an exploratory study on software which automatically recognizes and classifies basic emotional states (sadness, surprise, happiness, anger, fear and disgust). The study consists of generating and analyzing the graphs of speech signals with using: Pitch, intensity and formant properties of emotive speech. Indeed, facial feature extraction phase uses the mathematical formulation and measures a set of Action Units (AUs) for emotion classification. The efficiency of the methodology was evaluated by experimental tests on 300 individuals (150 females and 150 males, 20 to 48 years old) multi-ethnic groups, namely: (i) European, (ii) Asian Middle East and (iii) American. In the light of the experiments, the accuracy of the proposed model for emotional detection time was calculated 2.53 s as we defined primarily more distinct boundaries between emotions to classify features in a set of basic emotions.File | Dimensione | Formato | |
---|---|---|---|
ICROM2016.pdf
non disponibili
Descrizione: pdf editoriale
Tipologia:
Versione Editoriale (PDF)
Licenza:
Non pubblico
Dimensione
1.23 MB
Formato
Adobe PDF
|
1.23 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.