A system for recognition of emotions basedon speech analysis can have interesting applications inhuman-robot interaction. In this paper, we carry out anexploratory study on the possibility to use a proposedmethodology to recognize basic emotions (sadness, surprise,happiness, anger, fear and disgust) based on phoneticand acoustic properties of emotive speech with theminimal use of signal processing algorithms. We set upan experimental test, consisting of choosing three typesof speakers, namely: (i) ve adult European speakers, (ii)ve Asian (Middle East) adult speakers and (iii) ve adultAmerican speakers. The speakers had to repeat 6 sentencesin English (with durations typically between 1 s and3 s) in order to emphasize rising-falling intonation andpitch movement. Intensity, peak and range of pitch andspeech rate have been evaluated. The proposed methodologyconsists of generating and analyzing a graph of formant,pitch and intensity, using the open-source PRAATprogram. From the experimental results, it was possible torecognize the basic emotions in most of the cases.
展开▼
