Speech Based Analysis of Group Interactions : University of Maryland - College Park , August 27 , 2024
Speech Based Analysis of Group Interactions : University of Maryland - College Park , August 27 , 2024
From the report: “Human communication includes nonverbal elements of speech, such as intonation, pitch, loudness, turn-taking and speaking rate. While there is still a gap between analysis conducted with the methodologies of social sciences and automatic analysis approaches, these latter hold a great deal of promise in drawing connections between a physical layer(the measurable characteristics of speech signals) and an inferential layer (social and psychological information that the speech signals are expected to convey). This chapter describes the technology for speech-based analysis such as data collection, speaker diarization, paralanguage analysis, and computational analysis of constructs. Challenges include obtaining appropriate datasets and ensuring construct validity. We cover the state-of-the-art of current analysis, touching on specific efforts such as the detection of valence, roles, personality, and group cohesion. Future work in this area should attend to limitations, such as of measurement and ecological validity, and the ethics of using artificial intelligence techniques to characterize social and psychological phenomena.”
Authors - Fringi, Evangelia, Paletz, Susannah B.F., Vinciarelli, Alessandro
Subjects
Authors
Publishers
Format
Related Resources
