Multivariate Statistical Analysis and One-Pass Vector Quantization

Current speaker authentication algorithms are largely based on multivariate statistical theory. In this chapter, we introduce the most important technical components and concepts of multivariate analysis as they apply to speaker authentication: the multivariate Gaussian (also called normal) distribution, principal component analysis (PCA), vector quantization (VQ), and segmental K-means. These fundamental techniques have been used for statistical pattern recognition and will be used in our further discussions throughout this book. Understanding the basic concepts of these techniques is essential for understanding and developing speaker authentication algorithms.

[1]  Robert M. Gray,et al.  Vector Quantizers and Predictive Quantizers for Gauss-Markov Sources , 1982, IEEE Trans. Commun..

[2]  Stephen G. Wilson,et al.  Magnitude/Phase Quantization of Independent Gaussian Variates , 1980, IEEE Trans. Commun..

[3]  John B. Thomas,et al.  Optimal circularly symmetric quantizers , 1982 .

[4]  Joel Max,et al.  Quantizing for minimum distortion , 1960, IRE Trans. Inf. Theory.

[5]  Leszek Rutkowski.  Vector Quantization for Image Compression , 2004 .

[6]  Aaron E. Rosenberg,et al.  Report: A vector quantization approach to speaker recognition , 1987, AT&T Technical Journal.

[7]  Qi Li,et al.  Synthesizing neural networks by sequential addition of hidden nodes , 1994, Proceedings of 1994 IEEE International Conference on Neural Networks (ICNN'94).

[8]  Alex Acero,et al.  Spoken Language Processing , 2001 .

[9]  Robert M. Gray,et al.  An Algorithm for Vector Quantizer Design , 1980, IEEE Trans. Commun..

[10]  M. Paez,et al.  Minimum Mean-Squared-Error Quantization in Speech PCM and DPCM Systems , 1972, IEEE Trans. Commun..

[11]  J. MacQueen.  Some methods for classification and analysis of multivariate observations , 1967 .

[12]  Lawrence R. Rabiner,et al.  A segmental k-means training procedure for connected word recognition , 1986, AT&T Technical Journal.

[13]  Qi Li,et al.  Fast training algorithms for large data sets with application to classification of multispectral images , 1994, Proceedings of 1994 28th Asilomar Conference on Signals, Systems and Computers.

[14]  Qi Li,et al.  One-pass vector quantizer design by sequential pruning of the training data , 1995, Proceedings., International Conference on Image Processing.

[15]  Biing-Hwang Juang,et al.  The segmental K-means algorithm for estimating parameters of hidden Markov models , 1990, IEEE Trans. Acoust. Speech Signal Process..

[16]  Qi Li,et al.  Principal feature classification , 1997, IEEE Trans. Neural Networks.

[17]  Thomas R. Fischer,et al.  Vector Quantizer Design for Memoryless Gaussian, Gamma, and Laplacian Sources , 1984, IEEE Trans. Commun..

[18]  R. W. Harris,et al.  A comparison of several vector quantization codebook generation approaches , 1993, IEEE Trans. Image Process..

[19]  Andrew J. Viterbi,et al.  Error bounds for convolutional codes and an asymptotically optimum decoding algorithm , 1967, IEEE Trans. Inf. Theory.

[20]  Qi Li,et al.  Improving discriminant neural network (DNN) design by the use of principal component analysis , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.

[21]  Biing-Hwang Juang,et al.  A vector quantization approach to speaker recognition , 1985, ICASSP '85. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[22]  Charles E. Heckler,et al.  Applied Multivariate Statistical Analysis , 2005, Technometrics.