FRAUG: A FRAME RATE BASED DATA AUGMENTATION METHOD FOR DEPRESSION DETECTION FROM SPEECH SIGNALS.
FRAUG: A FRAME RATE BASED DATA AUGMENTATION METHOD FOR DEPRESSION DETECTION FROM SPEECH SIGNALS.
复制标题
DOI：
10.1109/icassp43922.2022.9746307
复制
发表时间：
2022-05
期刊：
影响因子：
--
通讯作者：
Alwan, Abeer
中科院分区：
文献类型：
--
作者：
Ravi, Vijay;Wang, Jinhan;Flint, Jonathan;Alwan, Abeer
提出了一种用于语音信号中抑郁检测的数据增强方法。在特征提取过程中，通过改变帧宽度和帧移位参数来创建数据增强样本。与其他数据增强方法(如VTLP、基音扰动或速度扰动)不同，该方法不显式地改变声学参数，而是改变帧级特征的时频分辨率。使用两个不同的数据集、模型和输入声学特征对提出的方法进行了评估。对于DAIC-WOZ(英语)数据集，当使用DepAudioNet模型和MEL谱图作为输入时，所提出的方法与基线相比分别提高了5.97%(验证)和25.13%(测试)。以CNN为后端，MFCC为输入特征的x向量嵌入，对Converge(普通话)数据集的改进分别为9.32%(验证)和12.99%(测试)。基线系统不包含任何数据增强。此外，该方法的性能优于常用的数据增强方法，如噪声增强、VTLP、速度和音调扰动。所有的改善在统计上都是显著的。
In this paper, a data augmentation method is proposed for depression detection from speech signals. Samples for data augmentation were created by changing the frame-width and the frame-shift parameters during the feature extraction process. Unlike other data augmentation methods (such as VTLP, pitch perturbation, or speed perturbation), the proposed method does not explicitly change acoustic parameters but rather the time-frequency resolution of frame-level features. The proposed method was evaluated using two different datasets, models, and input acoustic features. For the DAIC-WOZ (English) dataset when using the DepAudioNet model and mel-Spectrograms as input, the proposed method resulted in an improvement of 5.97% (validation) and 25.13% (test) when compared to the baseline. The improvements for the CONVERGE (Mandarin) dataset when using the x-vector embeddings with CNN as the backend and MFCCs as input features were 9.32% (validation) and 12.99% (test). Baseline systems do not incorporate any data augmentation. Further, the proposed method outperformed commonly used data-augmentation methods such as noise augmentation, VTLP, Speed, and Pitch Perturbation. All improvements were statistically significant.
登录
查看更多内容
影响因子：
4.5
作者：
He, Lang;Cao, Cui
通讯作者：
Cao, Cui
影响因子：
3.2
作者：
Cummins, Nicholas;Scherer, Stefan;Quatieri, Thomas F.
通讯作者：
Quatieri, Thomas F.
影响因子：
6.9
作者：
Li, Y.;Shi, S.;Yang, F.;Gao, J.;Li, Youhui;Tao, M.;Wang, G.;Zhang, K.;Gao, C.;Liu, L.;Li, Kan;Li, Keqing;Liu, Y.;Wang, Xumei;Zhang, J.;Lv, L.;Wang, Xueyi;Chen, Q.;Hu, J.;Sun, L.;Shi, J.;Chen, Y.;Xie, D.;Flint, J.;Kendler, K. S.;Zhang, Z.
通讯作者：
Zhang, Z.
影响因子：
3
作者：
McNemar, Quinn
通讯作者：
McNemar, Quinn
DOI：
10.1016/s0140-6736(18)32278-5
发表时间：
2018-11-10
期刊：
Lancet (London, England)
影响因子：
--
作者：
GBD 2017 Population and Fertility Collaborators
通讯作者：
GBD 2017 Population and Fertility Collaborators