Depression is a frequent affective disorder, leading to a high impact on patients, their families and society. Depression diagnosis is limited by assessment methods that rely on patient-reported or clinician judgments of symptom severity. Recently, many researches showed that voice is an objective indicator for depressive diagnosis. In this paper, we investigate a sample of 111 subjects (38 healthy controls, 36 mild depressed patients and 37 severe depressed patients) through comparative analysis to explore the correlation between acoustic features and depression severity. We extract features as many as possible according to previous researches to create a large voice feature set. Then we employ some feature selection methods to form compact subsets on different tasks. Finally, we evaluate depressive disorder severity by these acoustic feature subsets. Results show that interview is a better choice than reading and picture description for depression assessment. Meanwhile, speech signal correlate to depression severity in a medium-level with statistically significant (p < 0.01).
CITATION STYLE
Liu, Z., Hu, B., Liu, F., Kang, H., Li, X., Yan, L., & Wang, T. (2016). Evaluation of depression severity in speech. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9919 LNAI, pp. 312–321). Springer Verlag. https://doi.org/10.1007/978-3-319-47103-7_31
Mendeley helps you to discover research relevant for your work.