Modeling of acoustic signal energies with a generalized Frank copula. A linguistic conjecture is reviewed

by García, J. E. and González-López, V. A.

In this paper was selected a generalized Frank copula to model the dependence between the energy on two frequency bands of the speech signal, coming from eight languages. Was developed an algorithm that uses maximum likelihood to choose the best fitting copula's parameters. Through Bootstrap, estimates the variability of the parameters for each language and computes confidence regions by means of Voronoi tesselations. A linguistic conjecture which claims that the languages are organized in three rhythmic classes, was confirmed by the Voronoi regions. Modeling with a uniparametric Frank copula, the different degrees of dependence between the energies was quantified.

The whole paper is available here: www.tandfonline.com/doi/abs/10.1080/03610926.2013.866248#.Ux3WPuddWWt

 

NeuroMat

The Research, Innovation and Dissemination Center for Neuromathematics is hosted by the University of São Paulo and funded by FAPESP (São Paulo Research Foundation).

 

User login

 

Contact

Address:
1010 Matão Street - Cidade Universitária - São Paulo - SP - Brasil. 05508-090. See map.

Phone:
55 11 3091-1717

General contact email:
neuromat@numec.prp.usp.br

Media inquiries email:
comunicacao@numec.prp.usp.br