Deep learning for Depression Recognition from Speech

• Published in: Mobile networks and applications Vol. 29; no. 4; pp. 1212 - 1227
• Main Authors: Tian, Han, Zhu, Zhang, Jing, Xu
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.08.2024; Springer Nature B.V
• Subjects: Algorithms; Artificial neural networks; Communications Engineering; Computer Communication Networks; Deep learning; Diagnosis; Electrical Engineering; Emotion recognition; Engineering; Feature extraction; Feature selection; IT in Business; Machine learning; Mental depression; Networks; Noise reduction; Principal components analysis; Speech; Speech recognition; Voice recognition; CNN; Depression recognition; speech; PCA
• ISSN: 1383-469X; 1572-8153
• DOI: 10.1007/s11036-022-02086-3

In recent years, depression has been widely concerned, which makes people depressed, even suicidal, causing serious adverse consequences. In this paper, a multi information joint decision algorithm model is established by means of emotion recognition. The model is used to analyze the representative data of the subjects, and to assist in diagnosis of whether the subjects have depression. The main work is as follows: On the basis of exploring the speech characteristics of people with depressive disorder, this paper conducts an in-depth study of speech assisted depression diagnosis based on the speech data in the DAIC-WOZ dataset. First, the speech information is preprocessed, including speech signal pre emphasis, framing windowing, endpoint detection, noise reduction, etc. Secondly, OpenSmile is used to extract the features of speech signals, and the speech features that the features can reflect are studied and analyzed in depth. Then feature selection is carried out based on the influence of speech features and feature combination on depression diagnosis. Then, principal component analysis is used to reduce the dimension of data features. Finally, the convolutional neural network is used to modeling, testing and result analysis showed that the voice based diagnosis of depression was as high as 87%.