Please wait a minute...
FITEE
Front. Inform. Technol. Electron. Eng.  2015, Vol. 16 Issue (5): 358-366    DOI: 10.1631/FITEE.1400323
    
基于无监督特征学习的语音情感识别方法
Zheng-wei Huang, Wen-tao Xue, Qi-rong Mao
Department of Computer Science and Communication Engineering, Jiangsu University, Zhenjiang 212013, China
Speech emotion recognition with unsupervised feature learning
Zheng-wei Huang, Wen-tao Xue, Qi-rong Mao
Department of Computer Science and Communication Engineering, Jiangsu University, Zhenjiang 212013, China
 全文: PDF 
摘要： 目的：语音情感识别是人机交互的关键技术之一。同时，良好的情感特征对语音情感识别系统性能具有极大影响。目前的语音情感特征主要通过手工设计方法提取，对于其是否能够很好地刻画情感特性以及是否存在最优情感特征集，相关研究者并没有达成公认。所以有必要对语音情感特征提取进行进一步深入研究。
创新点：提出一种基于数据驱动的无监督情感特征学习方法。该方法能够自动从无标注语音数据中学习产生与情感相关的特征映射函数，用于语音情感特征提取。
方法：采用三种无监督学习算法（K-均值聚类，稀疏自动编码器，稀疏受限玻尔兹曼机）从若干无标注语音块中学习产生与目标相关的特征提取器，继而对整个语音样本进行特征提取（卷积和池化），最后训练一个线性支持向量机对未知样本进行识别。同时对模型涉及的超参数（块大小和隐层结点数目）进行选择。
结论：相对于传统原始特征，学习产生的特征具有一定的稀疏性并且对说话人及其他扰动因素具有一定鲁棒性。实验结果表明，尺寸较大的块和数量较多的隐层结点有助于提升系统性能（图4、5）。
关键词： 语音情感识别;  无监督特征学习;  神经网络;  情感计算    
Abstract: Emotion-based features are critical for achieving high performance in a speech emotion recognition (SER) system. In general, it is difficult to develop these features due to the ambiguity of the ground-truth. In this paper, we apply several unsupervised feature learning algorithms (including K-means clustering, the sparse auto-encoder, and sparse restricted Boltzmann machines), which have promise for learning task-related features by using unlabeled data, to speech emotion recognition. We then evaluate the performance of the proposed approach and present a detailed analysis of the effect of two important factors in the model setup, the content window size and the number of hidden layer nodes. Experimental results show that larger content windows and more hidden nodes contribute to higher performance. We also show that the two-layer network cannot explicitly improve performance compared to a single-layer network.
Key words: Speech emotion recognition    Unsupervised feature learning    Neural network    Affect computing
收稿日期: 2014-09-16 出版日期: 2015-05-05
CLC:  TP391.4  
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章　　
Zheng-wei Huang
Wen-tao Xue
Qi-rong Mao

引用本文:

Zheng-wei Huang, Wen-tao Xue, Qi-rong Mao. Speech emotion recognition with unsupervised feature learning. Front. Inform. Technol. Electron. Eng., 2015, 16(5): 358-366.

链接本文:

http://www.zjujournals.com/xueshu/fitee/CN/10.1631/FITEE.1400323    或    http://www.zjujournals.com/xueshu/fitee/CN/Y2015/V16/I5/358

[1] Yu-jun Xiao, Wen-yuan Xu, Zhen-hua Jia, Zhuo-ran Ma, Dong-lian Qi. 一种非侵入式的基于功耗的可编程逻辑控制器异常检测方案[J]. Frontiers of Information Technology & Electronic Engineering, 2017, 18(4): 519-534.
[2] Muhammad Asif Zahoor Raja, Iftikhar Ahmad, Imtiaz Khan, Muhammed Ibrahem Syam, Abdul Majid Wazwaz. 用于解决非线性受电弓系统的启发式神经网络计算[J]. Frontiers of Information Technology & Electronic Engineering, 2017, 18(4): 464-484.
[3] Guang-hui Song, Xiao-gang Jin, Gen-lang Chen, Yan Nie. 基于两级层次特征学习的图像分类方法[J]. Front. Inform. Technol. Electron. Eng., 2016, 17(9): 897-906.
[4] Gurmanik Kaur, Ajat Shatru Arora, Vijender Kumar Jain. 基于体位特征使用混杂模型预测血压对于无支撑后背的反应[J]. Front. Inform. Technol. Electron. Eng., 2015, 16(6): 474-485.
[5] Ying Cai, Meng-long Yang, Jun Li. 基于深度卷积网络的多分类法在头部姿态估计中的应用[J]. Front. Inform. Technol. Electron. Eng., 2015, 16(11): 930-939.
[6] Fei-wei Qin, Lu-ye Li, Shu-ming Gao, Xiao-ling Yang, Xiang Chen. 用于三维CAD模型分类的深度学习方法[J]. Front. Inform. Technol. Electron. Eng., 2014, 15(2): 91-106.
[7] Xiao-hua Wang, Juan-juan Yu, Yao Huang, Hua Wang, Zhong-hua Miao. 线性脉冲系统的自适应动态规划方法[J]. Front. Inform. Technol. Electron. Eng., 2014, 15(1): 43-50.