Stress and Emotion Recognition Using Acoustic Speech Analysis
This chapter describes computational methods for an automatic recognition of stress levels and different types of speaker’s emotions expressed in natural, not acted speech. A range of different acoustic features are examined and compared with respect to t
- PDF / 517,943 Bytes
- 22 Pages / 439.37 x 666.142 pts Page_size
- 78 Downloads / 250 Views
Stress and Emotion Recognition Using Acoustic Speech Analysis Margaret Lech and Ling He
9.1 Stress and Emotion Both stress and emotion represent psycho-physiological states involving characteristic somatic and autonomic responses [38]. Stress is a psychological and biological term characterized by loss of ability to appropriately respond to difficult emotional and physical conditions that can be either real or imagined. Stress is characterized by subjective strain, dysfunctional physiological activity, and deterioration of performance [14, 67]. Typical stress symptoms include fast heart rate, increased adrenaline production, difficulty to cope with relatively simple tasks, feeling of strain and exhaustion and inability to concentrate. Stress may be induced by external factors such as workload, noise, vibration or sleep loss or by internal factors such as fatigue [63]. Existing stress detection and classification research uses most frequently stress categories based on different levels of difficulties that a given person has to deal with [63]. Emotion comprises of complex psychological and physiological phenomena including person’s state of mind and the way an individual interacts with the environment. Generally, emotion involves: physiological arousal, expressive behaviors, and conscious experience [19]. Emotion is closely related to the state called mood. Unlike emotion which is a short-term (minutes–hours) psychophysiological state, mood is a relatively long lasting emotional state (hours– weeks). In contrast to simple emotions, moods are less specific, less intense, and less likely to be triggered by a particular stimulus or event [50].
M. Lech (&) School of Electrical and Computer Engineering, RMIT University, Melbourne 3001, Australia e-mail: [email protected] L. He Department of Medical Informatics and Engineering, School of Electrical Engineering and Information, Sichuan University, Chengdu, China
M. Lech et al. (eds.), Mental Health Informatics, Studies in Computational Intelligence 491, DOI: 10.1007/978-3-642-38550-6_9, Springer-Verlag Berlin Heidelberg 2014
163
164
M. Lech and L. He
The importance of emotional state and stress level analysis has been well recognized as a key factor in the Mental Status Examination. Most common, prototypical emotion models postulate that humans evolved to experience a limited number of prototypical emotions (e.g. happiness, anger, fear, sadness, etc.). Each discrete emotion is assumed to be an effect of a specific pattern of peripheral physiological response associated with a dedicated central nervous system representation [29, 49]. The more complex emotional experiences can then be understood as an effect of weighted integration of different brain regions to generate a range of emotional valence. Dynamics of emotions and the individual abilities to cope with different emotional states provide important cues in the diagnosis of mental disorders such as depression, near suicidal state and schizophrenia. The Diagnostic and Statistical Manual Disorder criter
Data Loading...