语音模拟器-CTC:抑郁症检测的顺序建模与语音时间分类
Jinhan Wang1, Vijay Ravi1, Jonathan Flint2
1Department of Electrical and Computer Engineering, University of California, Los Angeles, California, 90095, USA.
Speech communication
|October 4, 2024
概括
这项研究介绍了Speechformer-CTC,这是一种用于发言中抑郁症检测的新框架. 它有效地模拟了语音段内的微妙抑郁模式,提高了英语和普通话数据集的准确性,而不需要详细的转录.
科学领域:
- 计算语言学计算语言学
- 精神病学是一个精神病学.
- 机器学习 机器学习
背景情况:
- 基于语音的抑郁症检测通常使用简单的分类,忽略了语音中的微妙抑郁模式.
- 由于抑郁症指标在语音段中的分布不均,现有的方法难以通用.
- 在抑郁体中缺乏细粒度标签,阻碍了追踪动态抑郁模式.
研究的目的:
- 提出一种新的框架,Speechformer-CTC,用于模拟语音段中非均分布的抑郁特征.
- 通过结合连接主义时间分类 (CTC) 目标来解决传统语音分类的局限性.
- 评估框架与自动语音识别 (ASR) 功能的兼容性及其在各种数据集上的性能.
主要方法:
- 开发了使用连接主义时间分类 (CTC) 目标函数的语音形式-CTC框架.
- 引入了两个新的CTC标签生成策略:预期-热点和HUBERT.
- 整合了各种细粒度的框架,并测试了与ASR特性 (HuBERT,Whisper) 的兼容性.
主要成果:
- 在DAIC-WOZ (英语) 和CONVERGE (汉语) 数据集上实现了改善的抑郁症检测宏F1分数.
- 拥有HUBERT ASR功能和HUBERT政策的系统在DAIC-WOZ上获得了83.15%的F1评分,接近最先进的状态.
- 在使用Whisper功能和HUBERT策略的CONVERGE数据集上观察到显著的F1得分改善 (CONVERGE1上9.82%,CONVERGE2上18.47%).
结论:
- 在语音中模拟非均分布的抑郁模式显著有利于抑郁症检测准确度.
- 语音模拟器-CTC框架提供了一个强大且兼容的方法,即使使用标准的ASR功能.
- 拟议的方法可以识别语音中显著的抑郁区域,推进自动抑郁症检测系统.
相关概念视频
01:30Classification of Signals
417
In signal processing, signals are classified based on various characteristics: continuous-time versus discrete-time, periodic versus aperiodic, analog versus digital, and causal versus noncausal. Each category highlights distinct properties crucial for understanding and manipulating signals.
A continuous-time signal holds a value at every instant in time, representing information seamlessly. In contrast, a discrete-time signal holds values only at specific moments, often denoted as x(n), where...
A continuous-time signal holds a value at every instant in time, representing information seamlessly. In contrast, a discrete-time signal holds values only at specific moments, often denoted as x(n), where...
417
01:25Cognitive Therapy
146
Cognitive therapy, pioneered by Aaron T. Beck in the 1960s, is a structured approach to addressing psychological distress by focusing on the influence of thoughts on emotions and behaviors. All cognitive therapies involve the basic assumption that human beings have control over their feelings, and that how individuals feel about something depends on how they think about it. Unlike psychoanalytic methods that delve into unconscious processes or humanistic approaches emphasizing...
146