@ARTICLE{26583223_581962394_2022, 
	author = {Antonina Rafikova and Ekaterina Valueva and Anastasia Panfilova}, 
	keywords = {, acoustic speech feature, Big Five, personality traits, emotions, automatic personality recognitionneural networks},
	title = {Voice and Psychological Characteristics: A Contemporary Review},
	journal = {Psychology. Journal of Higher School of Economics},
	year = {2022},
	volume = {19},
	number = {1},
	pages = {195-215},
	url = {https://psy-journal.hse.ru/en/2022-19-1/581962394.html},
	publisher = {},
	abstract = {This paper provides an overview of studies examining the relationship between psychological characteristics and properties of voice. The article presents a historical overview of studies in this field, the main audio characteristics of speech used for voice analysis in modern researches (pitch, intensity and speed), research data on the acoustic and prosodic characteristics of emotions, intentions and personality traits (leadership qualities, charisma, Big Five personality traits). It is shown that negative emotions are recognized by voice better than positive ones, anger and sadness are most accurately recognized. The audio correlates of such measurements of emotions as valence and activation are considered. Acoustic-prosodic characteristics are described not only for basic emotions (anger, happiness, sadness, fear, disgust, surprise), but also of emotions that are of a less intense nature - irritation, resignation, indifference. Just like emotional speech, nonverbal vocalizations have a combination of acoustic properties, which can also be predicted based on their physical properties. It is shown that the perception of charismatic speech is due to a combination of prosodic and lexical properties of speech, and the role of auditory characteristics in the recognition of charisma is lower than in the recognition of emotions, where voice characteristics play a more prominent role than lexical content. When evaluating the Big Five traits using expert assessments, extraversion was found to have the largest number of significant correlations with acoustic-prosodic and auditory traits. Automatic personality and emotion recognition (using machine learning and neural networks) is a research area, where we can see the burst of empirical studies. Computer technologies assure a high degree of accuracy in personality prediction, but such studies are rarely deeply theoretically grounded. In conclusion, the need for theoretical understanding of the empirical results obtained is emphasized.},
	annote = {This paper provides an overview of studies examining the relationship between psychological characteristics and properties of voice. The article presents a historical overview of studies in this field, the main audio characteristics of speech used for voice analysis in modern researches (pitch, intensity and speed), research data on the acoustic and prosodic characteristics of emotions, intentions and personality traits (leadership qualities, charisma, Big Five personality traits). It is shown that negative emotions are recognized by voice better than positive ones, anger and sadness are most accurately recognized. The audio correlates of such measurements of emotions as valence and activation are considered. Acoustic-prosodic characteristics are described not only for basic emotions (anger, happiness, sadness, fear, disgust, surprise), but also of emotions that are of a less intense nature - irritation, resignation, indifference. Just like emotional speech, nonverbal vocalizations have a combination of acoustic properties, which can also be predicted based on their physical properties. It is shown that the perception of charismatic speech is due to a combination of prosodic and lexical properties of speech, and the role of auditory characteristics in the recognition of charisma is lower than in the recognition of emotions, where voice characteristics play a more prominent role than lexical content. When evaluating the Big Five traits using expert assessments, extraversion was found to have the largest number of significant correlations with acoustic-prosodic and auditory traits. Automatic personality and emotion recognition (using machine learning and neural networks) is a research area, where we can see the burst of empirical studies. Computer technologies assure a high degree of accuracy in personality prediction, but such studies are rarely deeply theoretically grounded. In conclusion, the need for theoretical understanding of the empirical results obtained is emphasized.}
}