Class-Dependent PCA Optimization Using Genetic Programming for Robust MFCC Extraction

Principal component analysis (PCA) is commonly used in feature extraction. It projects the features in direction of maximum variance. This projection can be performed in a class-dependent or class-independent manner. In this paper, we propose to optimize class-dependent PCA transformation matrix for robust MFCC feature extraction using genetic programming. For this purpose, we first map logarithm of clean speech Mel filter bank energies (LMFE) in directions of maximum variability. We obtain the mapping functions using genetic programming. After this, we form class-dependent PCA transformation matrix based on mapped LMFE and use this matrix in place of DCT in MFCC feature extraction. The experimental results show that proposed method achieves to significant isolated word recognition rate on Aurora2 database.

[1]  David Pearce,et al.  The aurora experimental framework for the performance evaluation of speech recognition systems under noisy conditions , 2000, INTERSPEECH.

[2]  E. C. Malthouse,et al.  Limitations of nonlinear PCA as performed with generic neural networks , 1998, IEEE Trans. Neural Networks.

[3]  Heng Tao Shen,et al.  Principal Component Analysis , 2009, Encyclopedia of Biometrics.

[4]  Ahmad Akbari,et al.  A framework for robust MFCC feature extraction using SNR-dependent compression of enhanced mel filter bank energies , 2006, INTERSPEECH.

[5]  Kuldip K. Paliwal,et al.  Class-dependent PCA, MDC and LDA: A combined classifier for pattern classification , 2006, Pattern Recognit..

[6]  Satoshi Nakamura,et al.  Maximum likelihood sub-band adaptation for robust speech recognition , 2005, Speech Commun..

[7]  Kohji Fukunaga,et al.  Introduction to Statistical Pattern Recognition-Second Edition , 1990 .

[8]  Heiga Zen,et al.  On the Use of Kernel PCA for Feature Extraction in Speech Recognition , 2003, IEICE Trans. Inf. Syst..

[9]  John R. Koza,et al.  Genetic programming - on the programming of computers by means of natural selection , 1993, Complex adaptive systems.

[10]  Panu Somervuo,et al.  Experiments with linear and nonlinear feature transformations in HMM based phone recognition , 2003, 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03)..

[11]  Nikos Fakotakis,et al.  Spectral and cepstral projection bases constructed by independent component analysis , 2000, INTERSPEECH.

[12]  Yung-Hwan Oh,et al.  On the use of channel-attentive MFCC for robust recognition of partially corrupted speech , 2004, IEEE Signal Process. Lett..

[13]  A. E. Eiben,et al.  Introduction to Evolutionary Computing , 2003, Natural Computing Series.

[14]  Alexander J. Smola,et al.  Learning with kernels , 1998 .

[15]  Bernhard Schölkopf,et al.  Nonlinear Component Analysis as a Kernel Eigenvalue Problem , 1998, Neural Computation.

[16]  Tetsuya Takiguchi,et al.  Robust Feature Extraction using Kernel PCA , 2006, 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.

[17]  M. Kramer.  Nonlinear principal component analysis using autoassociative neural networks , 1991 .