Segmentation natural speech using fractal dimension

Authors:
F. Martinez;A. Guillamon;J. J. Martinez
Affiliations:
Departamento de Matemática Aplicada y Estadistica, Universidad Politécnica de Cartagena, Spain;Departamento de Matemática Aplicada y Estadistica, Universidad Politécnica de Cartagena, Spain;Departamento de Matemática Aplicada y Estadistica, Universidad Politécnica de Cartagena, Spain
Venue:
ICCMSE '03 Proceedings of the international conference on Computational methods in sciences and engineering
Year:
2003

Citing 3
Cited 0

Approach to an irregular time series on the basis of the fractal theory

Physica D
Distance measures for signal processing and pattern recognition

Signal Processing
Kolmogorov Complexity of Finite Sequences and Recognition of Different Preictal EEG Patterns

CBMS '95 Proceedings of the Eighth Annual IEEE Symposium on Computer-Based Medical Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

Speech signals can be considered as being generated by mechanical systems with inherently nonlinear dynamics. The purpose of this paper is to present an automatic segmentation method based on nonlinear dynamics with low computational cost. The fractal dimension is a measure of signal complexity that can characterize different voiced and unvoiced. The segmentation process is carried out in two stages: estimation of the fractal dimension using the method suggested by Katz [1] and detection of the stationarity of the fractal dimension by means of the value of the variance parameter computed over the smooth fractal dimension signal. Using this combination of techniques, a quick and automatic segmentation is obtained. Our experiments have been computed over recorder signals from a speech Spanish database (AHUMADA).