Análise da autossimilaridade de sinais de voz baseada em wavelets na detecção de patologias laríngeas

Jayne dos Santos Lima, Stefanie G.Vilela, Washington C. de A. Costa, Silvana Cunha Costa, Suzete E. N. Correia

DOI: 10.14209/sbrt.2013.62

Evento: XXXI Simpósio Brasileiro de Telecomunicações (SBrT2013)

Keywords: Laryngeal pathology detection Hurst exponent Discrete Wavelet Transform Support Vector Machine

Abstract

This article employs the Hurst exponent, obtained by discrete wavelet transform, as a measure of self-similarity to characterize the nonlinearities of healthy voices and from those affected by vocal folds pathologies. To evaluate their discriminative potential, Support Vector Machines are used in the classification process. The results show self-similarity at thea nalyzed signals for the investigated wavelets.

Download