website menu

C4DM / Machine Listening Seminar : Andre Anikin

QMUL School of Electronic Engineering and Computer Science

Centre for Digital Music Seminar Series

Lecture by Andre Anikin, Lund University

Date/time:3-4pm, Tuesday, 4th December 2018

Location: ITL First floor seminar room Campus map:

Open to academics, students, alumni, public; all welcome. Admission is FREE, no pre-booking required.

Title: Human emotional vocalizations from an evolutionary perspective

Abstract: Emotional vocalizations have recently attracted considerable attention due to their abundance in human interactions, rich communicative potential, and significant cross-cultural similarities. A potential limitation is that most available recordings are actor portrayals. However, since some emotional expressions are known to be honest, hard-to-fake signals, volitional vocalizations may differ from spontaneous forms. In a series of studies we compiled and tested a large collection of naturalistic emotional vocalizations obtained from social media. The accuracy of emotion recognition was similar across cultures, suggesting that spontaneous vocalizations may be even less culture-specific than actor portrayals. In a comparison with several published corpora, listeners were above chance when guessing the spontaneous or volitional nature of vocalizations across all emotional categories, with accuracy highest for high-stakes emotions such as fear and anger. In addition, acoustic categories, such as “laugh” or “scream”, were more salient than emotional categories, suggesting that humans produce and perceive call types, while their interpretation in terms of emotion may be secondary. This brings the research on human non-lingustic vocalizations more in line with bioacoustics, which bodes well for the evolutionary approach and speaks in favor of using spontaneous expressions for the purposes of phylogenetic reconstruction.

Author info: Andre Anikin is a member of the Lund University Cognitive Science Group, Lund, Sweden.

Return to top