Learning Corpus-Invariant Discriminant Feature Representations for Speech Emotion Recognition

Peng SONG; Shifeng OU; Zhenbin DU; Yanyan GUO; Wenming MA; Jinglei LIU; Wenming ZHENG

doi:10.1587/transinf.2016EDL8222

Learning Corpus-Invariant Discriminant Feature Representations for Speech Emotion Recognition

Peng SONG, Shifeng OU, Zhenbin DU, Yanyan GUO, Wenming MA, Jinglei LIU, Wenming ZHENG

Full Text Views

0

Share
Cite this

Summary :

As a hot topic of speech signal processing, speech emotion recognition methods have been developed rapidly in recent years. Some satisfactory results have been achieved. However, it should be noted that most of these methods are trained and evaluated on the same corpus. In reality, the training data and testing data are often collected from different corpora, and the feature distributions of different datasets often follow different distributions. These discrepancies will greatly affect the recognition performance. To tackle this problem, a novel corpus-invariant discriminant feature representation algorithm, called transfer discriminant analysis (TDA), is presented for speech emotion recognition. The basic idea of TDA is to integrate the kernel LDA algorithm and the similarity measurement of distributions into one objective function. Experimental results under the cross-corpus conditions show that our proposed method can significantly improve the recognition rates.

Publication: IEICE TRANSACTIONS on Information Vol.E100-D No.5 pp.1136-1139

Publication Date: 2017/05/01

Publicized: 2017/02/02

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2016EDL8222

Type of Manuscript: LETTER

Category: Speech and Hearing

Authors

Peng SONG
  Yantai University
Shifeng OU
  Yantai University
Zhenbin DU
  Yantai University
Yanyan GUO
  Yantai University
Wenming MA
  Yantai University
Jinglei LIU
  Yantai University
Wenming ZHENG
  Southeast University

Keyword

speech emotion recognition, transfer learning, dimensionality reduction

Cite this

Copy

Peng SONG, Shifeng OU, Zhenbin DU, Yanyan GUO, Wenming MA, Jinglei LIU, Wenming ZHENG, "Learning Corpus-Invariant Discriminant Feature Representations for Speech Emotion Recognition" in IEICE TRANSACTIONS on Information, vol. E100-D, no. 5, pp. 1136-1139, May 2017, doi: 10.1587/transinf.2016EDL8222.
Abstract: As a hot topic of speech signal processing, speech emotion recognition methods have been developed rapidly in recent years. Some satisfactory results have been achieved. However, it should be noted that most of these methods are trained and evaluated on the same corpus. In reality, the training data and testing data are often collected from different corpora, and the feature distributions of different datasets often follow different distributions. These discrepancies will greatly affect the recognition performance. To tackle this problem, a novel corpus-invariant discriminant feature representation algorithm, called transfer discriminant analysis (TDA), is presented for speech emotion recognition. The basic idea of TDA is to integrate the kernel LDA algorithm and the similarity measurement of distributions into one objective function. Experimental results under the cross-corpus conditions show that our proposed method can significantly improve the recognition rates.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2016EDL8222/_p

Copy

@ARTICLE{e100-d_5_1136,
author={Peng SONG, Shifeng OU, Zhenbin DU, Yanyan GUO, Wenming MA, Jinglei LIU, Wenming ZHENG, },
journal={IEICE TRANSACTIONS on Information},
title={Learning Corpus-Invariant Discriminant Feature Representations for Speech Emotion Recognition},
year={2017},
volume={E100-D},
number={5},
pages={1136-1139},
abstract={As a hot topic of speech signal processing, speech emotion recognition methods have been developed rapidly in recent years. Some satisfactory results have been achieved. However, it should be noted that most of these methods are trained and evaluated on the same corpus. In reality, the training data and testing data are often collected from different corpora, and the feature distributions of different datasets often follow different distributions. These discrepancies will greatly affect the recognition performance. To tackle this problem, a novel corpus-invariant discriminant feature representation algorithm, called transfer discriminant analysis (TDA), is presented for speech emotion recognition. The basic idea of TDA is to integrate the kernel LDA algorithm and the similarity measurement of distributions into one objective function. Experimental results under the cross-corpus conditions show that our proposed method can significantly improve the recognition rates.},
keywords={},
doi={10.1587/transinf.2016EDL8222},
ISSN={1745-1361},
month={May},}

Copy

TY - JOUR
TI - Learning Corpus-Invariant Discriminant Feature Representations for Speech Emotion Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 1136
EP - 1139
AU - Peng SONG
AU - Shifeng OU
AU - Zhenbin DU
AU - Yanyan GUO
AU - Wenming MA
AU - Jinglei LIU
AU - Wenming ZHENG
PY - 2017
DO - 10.1587/transinf.2016EDL8222
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E100-D
IS - 5
JA - IEICE TRANSACTIONS on Information
Y1 - May 2017
AB - As a hot topic of speech signal processing, speech emotion recognition methods have been developed rapidly in recent years. Some satisfactory results have been achieved. However, it should be noted that most of these methods are trained and evaluated on the same corpus. In reality, the training data and testing data are often collected from different corpora, and the feature distributions of different datasets often follow different distributions. These discrepancies will greatly affect the recognition performance. To tackle this problem, a novel corpus-invariant discriminant feature representation algorithm, called transfer discriminant analysis (TDA), is presented for speech emotion recognition. The basic idea of TDA is to integrate the kernel LDA algorithm and the similarity measurement of distributions into one objective function. Experimental results under the cross-corpus conditions show that our proposed method can significantly improve the recognition rates.
ER -