Abstract
Emotion recognition from speech is an important field of research in human computer interaction. In this letter the framework of Support Vector Machines (SVM) with Gaussian Mixture Model (GMM) supervector is introduced for emotional speech recognition. Because of the importance of variance in reflecting the distribution of speech, the normalized mean vectors potential to exploit the information from the variance are adopted to form the GMM supervector. Comparative experiments from five aspects are conducted to study their corresponding effect to system performance. The experiment results, which indicate that the influence of number of mixtures is strong as well as influence of duration is weak, provide basis for the train set selection of Universal Background Model (UBM).
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
D. A. Reynolds. Speaker verification using adapted Gaussian mixture models. Digital Signal Processing, 10(2000)1, 19–41.
N. Cristianini, and J. S. Taylor. Support Vector Machines. Cambridge, Cambridge University Press, 2000, 125–126.
V. Wan and S. Renals. SVMSVM: support vector machine speaker verification methodology. Proceedings of International Conference on Acoustic, Speech and Signal Processing (ICASSP), Hong Kong, China, 2003, 221–224.
M. H. Liu, B. Q. Dai, Y. L. Xie, and Z. Q. Yao. Improved GMM-UBM/SVM for speaker verification. Proceedings of International Conference on Acoustic, Speech and Signal Processing (ICASSP), Toulouse, France, 2006, 96–99.
C. H. You, K. A. Lee, and H. Li. An SVM kernel with GMM-supervector based on the Bhattacharyya distance for speaker recognition. IEEE Signal Processing Letters, 16(2009)1, 49–52.
W. M. Campbell, D. E. Sturim, and D. A. Reynolds. Support vectormachines using GMM supervectors for speaker verification. IEEE Signal Processing Letters, 13(2006)5, 308–311.
W. M. Campbell, D. E. Sturim, D. A. Reynolds, and A. Solomonoff. SVM based speaker verification using a GMM supervector kernel and NAP variability compensation. Proceedings of International Conference on Acoustic, Speech and Signal Processing (ICASSP), Toulouse, France, 2006, 432–435.
Monto Carlo method. http://en.wikipedia.org/wiki/Monte_Carlo_method. 2012. 4.
F. Schwenker, S. Scherer, Y. M. Magdi, and G. Palm. The GMM-SVM supervector approach for the recognition of the emotional status from speech. Proceedings of International Conference on Artificial Neural Network (ICANN), Part I, Limassol, Cyprus, 2009, 894–903.
P. J. Moreno, P. P. Ho, and N. Vasconcelos. A Kullback-Leibler divergence based kernel for SVM classification in multimedia applications. Advances in Neural Information Processing Systems, Cambridge, MA, USA, MIT Press, 2004, 1–9.
Author information
Authors and Affiliations
Corresponding author
Additional information
Supported by the National Natural Science Foundation of China (No. 61105076), Natural Science Foundation of Anhui Province of China (No. 11040606M127) as well as Key Scientific and Technological Project of Anhui Province (No. 11010202192).
Communication author: Chen Yanxiang, born in 1972, female, Associate Professor.
About this article
Cite this article
Chen, Y., Xie, J. Emotional speech recognition based on SVM with GMM supervector. J. Electron.(China) 29, 339–344 (2012). https://doi.org/10.1007/s11767-012-0871-2
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11767-012-0871-2
Key words
- Emotional speech recognition
- Support Vector Machines (SVM)
- Gaussian Mixture Model (GMM) supervector
- Universal Background Model (USB)