Fifth International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2007)

Florence, Italy
December 13-15, 2007

Visualization of Normal and Pathological Speech Data

J. C. Goddard (1), F. M. Martínez (1), G. Schlotthauer (2), M. E. Torres (2), H. L. Rufiner (2,3)

(1) Departamento de Ingeniería Eléctrica, Universidad Autonoma Metropolitana, Iztapalapa, Mexico
(2) Facultad de Ingeniería, Universidad Nacional de Entre Ríos, Paraná, Argentina
(3) Facultad de Ingeniería y Ciencias Hídricas, Universidad Nacional de Litoral, Santa Fe, Argentina

Techniques for the visualization of highdimensional data are common in exploratory data analysis and can be very useful for gaining an intuition into the structure of a data set. The classical method of principal component analysis is the one most often employed, however in recent years a number of other nonlinear techniques have been introduced. In the present paper, principal component analysis, and two newer methods, are applied to a set of speech data and their results are compared.
Index Terms. PCA, LLE, Kernel PCA

Full Paper (reprinted with permission from Firenze University Press)

Bibliographic reference.  Goddard, J. C. / Martínez, F. M. / Schlotthauer, G. / Torres, M. E. / Rufiner, H. L. (2007): "Visualization of normal and pathological speech data", In MAVEBA-2007, 33-36.