Depressive Corpus Construction of Feature Selection and Analysis

Depression is a serious mental disorder. The diagnosis method is mainly based on subjective scales and lacks objective indicators. Speech, as a non-invasive, easy-to-collect, low-cost objective indicator, has attracted the attention of researchers. There are many public depression corpora in the world, but due to cultural and language differences, there is no publicly available Chinese depression corpus. At the same time, there are still no speech acoustic features that can fully and effectively identify depression. Based on the theory of abnormal self-processing in psychology, this study designed a Chinese depression text corpus, collected the voices of 60 subjects, used experimental statistics methods to conduct significant comparative analysis and correlation analysis on the acoustic features, and finally concluded that there are effective acoustic signatures that differ between genders. The acoustic characteristics of depression have important application value in the field of mental health and also provide a reference for the establishment of a depression corpus.