An Investigation of Emotional Speech in Depression Classification

Brian Stasek, Julien Epps, Nicholas Cummins, Roland GOECKE

    Research output: A Conference proceeding or a Chapter in BookConference contribution

    12 Citations (Scopus)
    2 Downloads (Pure)

    Abstract

    Assessing depression via speech characteristics is a growing area of interest in quantitative mental health research with a view to a clinical mental health assessment tool. As a mood disorder, depression induces changes in response to emotional stimuli, which motivates this investigation into the relationship between emotion and depression affected speech. This paper investigates how emotional information expressed in speech (i.e. arousal, valence, dominance) contributes to the classification of minimally depressed and moderately-severely depressed individuals. Experiments based on a subset of the AVEC 2014 database show that manual emotion ratings alone are discriminative of depression and combining rating-based emotion features with acoustic features improves classification between mild and severe depression. Emotion-based data selection is also shown to provide improvements in depression classification and a range of threshold methods are explored. Finally, the experiments presented demonstrate that automatically predicted emotion ratings can be incorporated into a fully automatic depression classification to produce a 5% accuracy improvement over an acoustic-only baseline system.
    Original languageEnglish
    Title of host publicationProceedings of Interspeech 2016
    EditorsNelson Morgan
    Place of PublicationSan Francisco
    PublisherInternational Speech Communication Association
    Pages485-489
    Number of pages5
    ISBN (Print)9781510833135
    Publication statusPublished - 2016
    EventInterspeech 2016 - San Francisco, San Francisco, United States
    Duration: 8 Sep 201612 Sep 2016
    http://www.interspeech2016.org/ (Conference website)

    Conference

    ConferenceInterspeech 2016
    CountryUnited States
    CitySan Francisco
    Period8/09/1612/09/16
    OtherInterspeech 2016 will be organized around the topic: Understanding Speech Processing in Humans and Machines. The event will be held in the Hyatt Regency San Francisco hotel in the beautiful San Francisco, California. Interspeech 2016 emphasizes an interdisciplinary approach covering all aspects of speech science and technology spanning basic theories to applications. In addition to regular oral and poster sessions, the conference will also feature plenary talks by internationally renowned experts, tutorials, special sessions, show & tell sessions, and exhibits. A number of satellite events will take place immediately before and after the conference
    Internet address

    Fingerprint

    Depression
    Emotions
    Acoustics
    Mental Health
    Arousal
    Mood Disorders
    Databases
    Research

    Cite this

    Stasek, B., Epps, J., Cummins, N., & GOECKE, R. (2016). An Investigation of Emotional Speech in Depression Classification. In N. Morgan (Ed.), Proceedings of Interspeech 2016 (pp. 485-489). San Francisco: International Speech Communication Association.
    Stasek, Brian ; Epps, Julien ; Cummins, Nicholas ; GOECKE, Roland. / An Investigation of Emotional Speech in Depression Classification. Proceedings of Interspeech 2016. editor / Nelson Morgan. San Francisco : International Speech Communication Association, 2016. pp. 485-489
    @inproceedings{090d31e615ef45e4a3ea44fda0794ccb,
    title = "An Investigation of Emotional Speech in Depression Classification",
    abstract = "Assessing depression via speech characteristics is a growing area of interest in quantitative mental health research with a view to a clinical mental health assessment tool. As a mood disorder, depression induces changes in response to emotional stimuli, which motivates this investigation into the relationship between emotion and depression affected speech. This paper investigates how emotional information expressed in speech (i.e. arousal, valence, dominance) contributes to the classification of minimally depressed and moderately-severely depressed individuals. Experiments based on a subset of the AVEC 2014 database show that manual emotion ratings alone are discriminative of depression and combining rating-based emotion features with acoustic features improves classification between mild and severe depression. Emotion-based data selection is also shown to provide improvements in depression classification and a range of threshold methods are explored. Finally, the experiments presented demonstrate that automatically predicted emotion ratings can be incorporated into a fully automatic depression classification to produce a 5{\%} accuracy improvement over an acoustic-only baseline system.",
    keywords = "Emotional-speech-in-depression",
    author = "Brian Stasek and Julien Epps and Nicholas Cummins and Roland GOECKE",
    year = "2016",
    language = "English",
    isbn = "9781510833135",
    pages = "485--489",
    editor = "Nelson Morgan",
    booktitle = "Proceedings of Interspeech 2016",
    publisher = "International Speech Communication Association",

    }

    Stasek, B, Epps, J, Cummins, N & GOECKE, R 2016, An Investigation of Emotional Speech in Depression Classification. in N Morgan (ed.), Proceedings of Interspeech 2016. International Speech Communication Association, San Francisco, pp. 485-489, Interspeech 2016, San Francisco, United States, 8/09/16.

    An Investigation of Emotional Speech in Depression Classification. / Stasek, Brian; Epps, Julien; Cummins, Nicholas; GOECKE, Roland.

    Proceedings of Interspeech 2016. ed. / Nelson Morgan. San Francisco : International Speech Communication Association, 2016. p. 485-489.

    Research output: A Conference proceeding or a Chapter in BookConference contribution

    TY - GEN

    T1 - An Investigation of Emotional Speech in Depression Classification

    AU - Stasek, Brian

    AU - Epps, Julien

    AU - Cummins, Nicholas

    AU - GOECKE, Roland

    PY - 2016

    Y1 - 2016

    N2 - Assessing depression via speech characteristics is a growing area of interest in quantitative mental health research with a view to a clinical mental health assessment tool. As a mood disorder, depression induces changes in response to emotional stimuli, which motivates this investigation into the relationship between emotion and depression affected speech. This paper investigates how emotional information expressed in speech (i.e. arousal, valence, dominance) contributes to the classification of minimally depressed and moderately-severely depressed individuals. Experiments based on a subset of the AVEC 2014 database show that manual emotion ratings alone are discriminative of depression and combining rating-based emotion features with acoustic features improves classification between mild and severe depression. Emotion-based data selection is also shown to provide improvements in depression classification and a range of threshold methods are explored. Finally, the experiments presented demonstrate that automatically predicted emotion ratings can be incorporated into a fully automatic depression classification to produce a 5% accuracy improvement over an acoustic-only baseline system.

    AB - Assessing depression via speech characteristics is a growing area of interest in quantitative mental health research with a view to a clinical mental health assessment tool. As a mood disorder, depression induces changes in response to emotional stimuli, which motivates this investigation into the relationship between emotion and depression affected speech. This paper investigates how emotional information expressed in speech (i.e. arousal, valence, dominance) contributes to the classification of minimally depressed and moderately-severely depressed individuals. Experiments based on a subset of the AVEC 2014 database show that manual emotion ratings alone are discriminative of depression and combining rating-based emotion features with acoustic features improves classification between mild and severe depression. Emotion-based data selection is also shown to provide improvements in depression classification and a range of threshold methods are explored. Finally, the experiments presented demonstrate that automatically predicted emotion ratings can be incorporated into a fully automatic depression classification to produce a 5% accuracy improvement over an acoustic-only baseline system.

    KW - Emotional-speech-in-depression

    M3 - Conference contribution

    SN - 9781510833135

    SP - 485

    EP - 489

    BT - Proceedings of Interspeech 2016

    A2 - Morgan, Nelson

    PB - International Speech Communication Association

    CY - San Francisco

    ER -

    Stasek B, Epps J, Cummins N, GOECKE R. An Investigation of Emotional Speech in Depression Classification. In Morgan N, editor, Proceedings of Interspeech 2016. San Francisco: International Speech Communication Association. 2016. p. 485-489