Skip to Main content Skip to Navigation
Conference papers

First Experiments to Detect Anomaly Using Personality Traits vs. Prosodic Features

Cédric Fayet 1 Arnaud Delhay 1 Damien Lolive 1 Pierre-François Marteau 1
1 EXPRESSION - Expressiveness in Human Centered Data/Media
UBS - Université de Bretagne Sud, IRISA-D6 - MEDIA ET INTERACTIONS
Abstract : This paper presents the design of an anomaly detector based on three diff erent sets of features, one corresponding to some prosodic descriptors and two extracted from Big Five traits. Big Five traits correspond to a simple but efficient representation of a human personality. They are extracted from a manual annotation while prosodic features are extracted directly from the speech signal. We evaluate two di fferent anomaly detection methods: One-Class SVM (OC-SVM) and iForest, each one combined with a threshold classi ffication to decide the "normality" of a sample. The di fferent combinations of models and feature sets are evaluated on the SSPNET-Personality corpus which has already been used in several experiments, including a previous work on separating two types of personality pro les in a supervised way. In this work, we propose the above mentioned unsupervised methods, and discuss their performance, to detect particular audio-clips produced by a speaker with an abnormal personality. Results show that using automatically extracted prosodic features competes with the Big Five traits. In our case, OCSVM seems to get better results than iForest.
Complete list of metadata
Contributor : Expression Irisa Connect in order to contact the contributor
Submitted on : Thursday, September 7, 2017 - 3:04:14 PM
Last modification on : Tuesday, October 19, 2021 - 11:58:59 PM


  • HAL Id : hal-01583539, version 1


Cédric Fayet, Arnaud Delhay, Damien Lolive, Pierre-François Marteau. First Experiments to Detect Anomaly Using Personality Traits vs. Prosodic Features. 19th International Conference on Speech and Computer (SPECOM), Sep 2017, Hatfield, Hertfordshire, United Kingdom. ⟨hal-01583539⟩



Record views