Multi-class SVM for stressed speech recognition

被引：0

作者：

Besbes, Salsabil ^{[1
]}

Lachiri, Lied ^{[2
]}

机构：

[1] Univ Tunis El Manar, Natl Sch Engineers Tunis, Signal Image & Informat Technol Lab, BP 37 Le Belvdre, Tunis 1002, Tunisia

[2] Univ Tunis El Manar, Natl Sch Engineers Tunis, BP 37 Le Belvdre, Tunis 1002, Tunisia

来源：

2016 2ND INTERNATIONAL CONFERENCE ON ADVANCED TECHNOLOGIES FOR SIGNAL AND IMAGE PROCESSING (ATSIP) | 2016年

关键词：

speech recognition; multi-class support vector machines; stressed context; SUSAS database; GFCC;

D O I：

暂无

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

This paper deals with a new automatic stressed recognition system based on kernel classification. We extracted advanced acoustic features from the stressed signals and employed a multi-class Support Vector Machines with different kernels to recognize speech utterances under stress. Gammatone Frequency Cepstral Coefficients are also established. The system implemented is tested using isolated words from SUSAS database with 4 classes: Neutral, Angry, Lombard and Loud. Experimental results show that the best performance is obtained when we use the auditory feature with different descriptors combination but it depends on the type of the kernel used.

引用

页码：782 / 787

页数：6

共 50 条

[1] A Multi-Class SVM Based Phonemes Classifier Based on a Trainable Confidence Measure
Amini, Sahar
Razzazi, Farbod
Nayebi, Kambiz
2009 IEEE INTERNATIONAL SYMPOSIUM ON SIGNAL PROCESSING AND INFORMATION TECHNOLOGY (ISSPIT 2009), 2009, : 49 - +
[2] Body type recognition by multi-class support vector machines
Li, Xiaojiu
Liu, Hao
Wang, Yuxiu
Li, Xiaozhi
International Journal of Advancements in Computing Technology, 2012, 4 (20) : 487 - 496
[3] An expert diagnosis system for classification of human parasite eggs based on multi-class SVM
Avci, Derya
Varol, Asaf
EXPERT SYSTEMS WITH APPLICATIONS, 2009, 36 (01) : 43 - 48
[4] HMM/SVM segmentation and labelling of Arabic speech for speech recognition applications
Frihia H.
Bahi H.
International Journal of Speech Technology, 2017, 20 (3) : 563 - 573
[5] Stressed speech recognition using a warped frequency scale
Gharavian, D.
Ahadi, S. M.
IEICE ELECTRONICS EXPRESS, 2008, 5 (06) : 187 - 191
[6] AN SVM BASED CONFIDENCE MEASURE FOR CONTINUOUS SPEECH RECOGNITION
Bardideh, Mohsen
Razzazi, Farbod
Ghassemian, Hassan
ICSPC: 2007 IEEE INTERNATIONAL CONFERENCE ON SIGNAL PROCESSING AND COMMUNICATIONS, VOLS 1-3, PROCEEDINGS, 2007, : 1015 - +
[7] High resolution speech feature parametrization for monophone-based stressed speech recognition
Sarikaya, R
Hansen, JHL
IEEE SIGNAL PROCESSING LETTERS, 2000, 7 (07) : 182 - 185
[8] A Subspace Projection Based Approach to Improve the Recognition of Stressed Speech
Priya, Bhanu
Dandapat, S.
2016 IEEE ANNUAL INDIA CONFERENCE (INDICON), 2016,
[9] Classification of speech under stress based on cepstral features and One-class SVM
Besbes, Salsabil
Lachiri, Zied
2017 INTERNATIONAL CONFERENCE ON CONTROL, AUTOMATION AND DIAGNOSIS (ICCAD), 2017, : 213 - 218
[10] SVM-based combination of confidence measures in speech recognition
Department of Electronic Engineering, School of Information Science and Technology, Beijing Institute of Technology, Beijing 100081, China
Beijing Ligong Daxue Xuebao, 2007, 3 (255-259):

← 1 2 3 4 5 →