Emotion recognition by speech signal characteristics (linguistic, clinical, informative aspects)

The paper presents an experimental project of linguists, medical professionals, lawyers, com-puter security specialists dealing with emotions discernment by basic speech signal characteristics. The software has been created, and its first testing has been carried out in the social network VKontakte. The collected recordings of speech fragments of live spontaneous prox-imate-intermediated dialogical speech were analyzed at several levels. First, a complex lin-guistic analysis revealed lexico-semantic and prosodic features of emotionality. Then, a com-parison with the software results was carried out, and the data obtained were systematized. Also, conclusions on the leading role of prosody in revealing hidden types of emotional stress were made. Frequent agreements of digital meanings of prosodic elements in speech segments were found demonstrating emotions not fixed at the lexico-semantic level. Finally, the work-ing formulation for external and internal ways of emotionality expression in the live speech was offered.

Dialogical self · Emotionality · Linguistics · Natural language processing · Prosody · Speech recognition · Computer Science · Cultural, Linguistic, Economic Studies · Discourse Analysis and Cultural Communication · Language, Communication, and Linguistic Studies · Psychology · Social Psychology