An attempt to estimate depressive status from voice
An attempt to estimate depressive status from voice
复制标题
尝试从声音估计抑郁状态
DOI：
10.1007/978-3-030-25872-6_13
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
and Shinichi Tokuno
中科院分区：
文献类型：
--
作者：
Yasuhiro Omiya;Takeshi Takano;Tomotaka Uraguchi;Mitsuteru Nakamura;Masakazu Higuchi;Shuji Shinohara;Shunji Mitsuyoshi;Mirai So;and Shinichi Tokuno
在全世界特别是发达国家，日益增多的心理健康障碍是一个严重的问题。作为一种对策，本文的主要目的是试图从声音来估计抑郁状态。在这项研究中，我们在医院的咨询室收集了重度抑郁症患者。采用“汉密尔顿抑郁量表”（HAM-D）对患者进行问卷调查。声音对应于三个长元音记录从受试者。接下来，基于语音计算声学特征量。我们开发了HAM-D分数估计算法，从语音使用三种类型的长元音音频内容之一。结果，“实际HAM-D评分”和“估计HAM-D评分”之间存在相关性。我们发现，该算法是有效的估计抑郁症的状态，并可以用于估计基于语音的疾病状态。
In the whole world especially developed countries, increasing mental health disorders is a serious problem. As a countermeasure, the main objective of this paper is an attempt to estimate depressive status from voice. In this study, we gathered patients with major depressive disorders in the hospital’s consulting room. Several questionnaires including “the Hamilton Depression Rating Scale” (HAM-D) were administered to evaluate the patients’ depressed state. Voices corresponding to three long vowels were recorded from the subjects. Next, the acoustic feature quantity was calculated based on the voice. We developed the HAM-D score estimation algorithm from the voice using one of three types of long vowel audio content. As a result, there was a correlation between the “Actual HAM-D Score” and the “Estimated HAM-D Score”. We found that the algorithm is effective in estimating depression state and can be used for estimating the disease state based on voice.