Mental Health Detection from Speech Signal: A Convolution Neural Networks Approach

EasyChair Preprint 2371

Abstract

Depression has long been recognized as one of the leading causes of disability and burden worldwide. In psychology, it is well known that the self is not only the cognitive subject, but also the core of personality. And the high incidence of suicide and pervasive hopelessness in depressed individuals suggested that the self might be abnormal among them. In light of the psychological characteristics, we employ classical scientific psychology paradigms on abnormalities of self-related processing in depressed individuals to develop a Chinese depressed speech corpus. Eleven depressed individuals and ten healthy subjects, who are gender-balanced and age-balanced, were recruited to participate in this work. Currently we have preliminarily collected 6 and 2.5 hours of speech data respectively, with the results of preliminary analysis indicating that there exist abnormalities in the depressed speech. The study results will provide a new perspective and strategy for further study on the building and application of speech corpus in depression

Keyphrases: Convolutional Neural Network, Depression, speech signal

@booklet{EasyChair:2371,
  author    = {Haizhen An and Xiaoyong Lu and Renjun Li and Daimin Shi and Jingyi Yuan and Tao Pan},
  title     = {Mental Health Detection from Speech Signal: A Convolution Neural Networks Approach},
  howpublished = {EasyChair Preprint 2371},
  year      = {EasyChair, 2020}}