Pervasive Computing Paradigms for Mental Health. Selected Papers from MindCare 2016, Fabulous 2016, and IIoT 2015

Research Article

Stress Detection from Speech Using Spectral Slope Measurements

  • @INPROCEEDINGS{10.1007/978-3-319-74935-8_5,
        author={Olympia Simantiraki and Giorgos Giannakakis and Anastasia Pampouchidou and Manolis Tsiknakis},
        title={Stress Detection from Speech Using Spectral Slope Measurements},
        proceedings={Pervasive Computing Paradigms for Mental Health. Selected Papers from MindCare 2016, Fabulous 2016, and IIoT 2015},
        proceedings_a={MINDCARE \& IIOT \& FABULOUS},
        year={2018},
        month={3},
        keywords={Stress detection Speech analysis Glottal source Fundamental frequency Spectral tilt Iterative adaptive inverse filtering Random forests},
        doi={10.1007/978-3-319-74935-8_5}
    }
    
  • Olympia Simantiraki
    Giorgos Giannakakis
    Anastasia Pampouchidou
    Manolis Tsiknakis
    Year: 2018
    Stress Detection from Speech Using Spectral Slope Measurements
    MINDCARE & IIOT & FABULOUS
    Springer
    DOI: 10.1007/978-3-319-74935-8_5
Olympia Simantiraki1,*, Giorgos Giannakakis1, Anastasia Pampouchidou2, Manolis Tsiknakis,*
  • 1: Foundation for Research and Technology–Hellas (FORTH–ICS)
  • 2: University of Burgundy
*Contact email: osimantir@ics.forth.gr, tsiknaki@ics.forth.gr

Abstract

Automatic detection of emotional stress is an active research domain, which has recently drawn increasing attention, mainly in the fields of computer science, linguistics, and medicine. In this study, stress is automatically detected by employing speech-derived features. Related studies utilize features such as overall intensity, MFCCs, Teager Energy Operator, and pitch. The present study proposes a novel set of features based on the spectral tilt of the glottal source and of the speech signal itself. The proposed features rely on the Probability Density Function of the estimated spectral slopes, and consist of the three most probable slopes from the glottal source, as well as the corresponding three slopes of the speech signal, obtained on a word level. The performance of the proposed method is evaluated on the simulated dataset of the SUSAS corpus, achieving recognition accuracy of , when the Random Forests classifier is used.