For cross-corpus speech emotion recognition (SER), how to obtain effective feature representation for the discrepancy elimination of feature distributions between source and target domains is a crucial issue. In this paper, we propose a Target-adapted Subspace Learning (TaSL) method for cross-corpus SER. The TaSL method trys to find a projection subspace, where the feature regress the label more accurately and the gap of feature distributions in target and source domains is bridged effectively. Then, in order to obtain more optimal projection matrix, ℓ1 norm and ℓ2,1 norm penalty terms are added to different regularization terms, respectively. Finally, we conduct extensive experiments on three public corpuses, EmoDB, eNTERFACE and AFEW 4.0. The experimental results show that our proposed method can achieve better performance compared with the state-of-the-art methods in the cross-corpus SER tasks.
Xiuzhen CHEN
Nanjing University of Information Science and Technology
Xiaoyan ZHOU
Nanjing University of Information Science and Technology
Cheng LU
Southeast University
Yuan ZONG
Southeast University
Wenming ZHENG
Southeast University
Chuangao TANG
Southeast University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Xiuzhen CHEN, Xiaoyan ZHOU, Cheng LU, Yuan ZONG, Wenming ZHENG, Chuangao TANG, "Target-Adapted Subspace Learning for Cross-Corpus Speech Emotion Recognition" in IEICE TRANSACTIONS on Information,
vol. E102-D, no. 12, pp. 2632-2636, December 2019, doi: 10.1587/transinf.2019EDL8038.
Abstract: For cross-corpus speech emotion recognition (SER), how to obtain effective feature representation for the discrepancy elimination of feature distributions between source and target domains is a crucial issue. In this paper, we propose a Target-adapted Subspace Learning (TaSL) method for cross-corpus SER. The TaSL method trys to find a projection subspace, where the feature regress the label more accurately and the gap of feature distributions in target and source domains is bridged effectively. Then, in order to obtain more optimal projection matrix, ℓ1 norm and ℓ2,1 norm penalty terms are added to different regularization terms, respectively. Finally, we conduct extensive experiments on three public corpuses, EmoDB, eNTERFACE and AFEW 4.0. The experimental results show that our proposed method can achieve better performance compared with the state-of-the-art methods in the cross-corpus SER tasks.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019EDL8038/_p
Copy
@ARTICLE{e102-d_12_2632,
author={Xiuzhen CHEN, Xiaoyan ZHOU, Cheng LU, Yuan ZONG, Wenming ZHENG, Chuangao TANG, },
journal={IEICE TRANSACTIONS on Information},
title={Target-Adapted Subspace Learning for Cross-Corpus Speech Emotion Recognition},
year={2019},
volume={E102-D},
number={12},
pages={2632-2636},
abstract={For cross-corpus speech emotion recognition (SER), how to obtain effective feature representation for the discrepancy elimination of feature distributions between source and target domains is a crucial issue. In this paper, we propose a Target-adapted Subspace Learning (TaSL) method for cross-corpus SER. The TaSL method trys to find a projection subspace, where the feature regress the label more accurately and the gap of feature distributions in target and source domains is bridged effectively. Then, in order to obtain more optimal projection matrix, ℓ1 norm and ℓ2,1 norm penalty terms are added to different regularization terms, respectively. Finally, we conduct extensive experiments on three public corpuses, EmoDB, eNTERFACE and AFEW 4.0. The experimental results show that our proposed method can achieve better performance compared with the state-of-the-art methods in the cross-corpus SER tasks.},
keywords={},
doi={10.1587/transinf.2019EDL8038},
ISSN={1745-1361},
month={December},}
Copy
TY - JOUR
TI - Target-Adapted Subspace Learning for Cross-Corpus Speech Emotion Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 2632
EP - 2636
AU - Xiuzhen CHEN
AU - Xiaoyan ZHOU
AU - Cheng LU
AU - Yuan ZONG
AU - Wenming ZHENG
AU - Chuangao TANG
PY - 2019
DO - 10.1587/transinf.2019EDL8038
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E102-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2019
AB - For cross-corpus speech emotion recognition (SER), how to obtain effective feature representation for the discrepancy elimination of feature distributions between source and target domains is a crucial issue. In this paper, we propose a Target-adapted Subspace Learning (TaSL) method for cross-corpus SER. The TaSL method trys to find a projection subspace, where the feature regress the label more accurately and the gap of feature distributions in target and source domains is bridged effectively. Then, in order to obtain more optimal projection matrix, ℓ1 norm and ℓ2,1 norm penalty terms are added to different regularization terms, respectively. Finally, we conduct extensive experiments on three public corpuses, EmoDB, eNTERFACE and AFEW 4.0. The experimental results show that our proposed method can achieve better performance compared with the state-of-the-art methods in the cross-corpus SER tasks.
ER -