Artificial intelligence

Abstract: Modern approaches to automated recognition of emotions and psychological conditions by voice are described. The structure system for speaker emotion identification that uses a preprocessing audio signal (noise reduction and segmentation by participants) and a set of acoustic and prosodic features of speech and extra linguistic feature to create describing vector are proposed. The results of numeric research point to the necessity to use these characteristics together.

Keywords: speaker model, acoustic characteristics of emotions, the Gaussian mixture method

References:

1. VoiceStress AnalysisServices// [El. recource]. – Access mode: http://www.voicestressanalysis.net
2. Potapova R. K. O vozmozhnosti pertseptivno-sluhovogo raspoznavaniya sostoyaniya «agressiya» po ustnoy rechi /Potapova R. K., Komalova L. R. // Vestnik MGLU. – 2014. – #13 (699) . –S.202-214.
3. Brose A. Affective states contribute to trait reports of affective well-being / Brose A., Lindenberg U.,Schmiedek F. // Emition .- 2013(5) .- p. 940-948.
4. Kächele M. Prosodic, Spectral and Voice Quality Feature Selection Using a Long-Term Stopping Criterion for Audio-Based Emotion Recognition / M. Kächele, D. Zharkov, S. Meudt and F. Schwenker // Pattern Recognition (ICPR), 2014 22nd International Conference on,Stockholm . – 2014. – pp. 803-808.
5. Klimenko N. S. Segmentatsiya i klassifikatsiya uchastkov rechevogo signala dlya postroeniya modeley diktorov v sisteme identifikatsii govoryaschego // Materialyi Mezhdunarodnoy nauchno-tehnich. konf. "Iskusstvennyiy intellekt. Intellektualnyie sistemyi". - Donetsk: IPShI "Nauka I osvIta", 2012. -S. 80-82.
6. Psihologiya emotsiy / Izard K.E.Perev.s angl. –SPb: Izdatelstvo «Piter», 1999. – 464 s.
7. Chernyaev L.S. Vzglyad geshtalt-terapevta na bazovyie emotsii / L.S. Chernyaev // Geshtalt 2007. – Ch 2. – Filosofiya i etika v geshtalt-podhode. – M.: MGI, 2007. –S.17-24.