Group delay function followed by dynamic programming versus multiscale-product for glottal closure instant detection

Abstract

This paper fits in the context of glottal closure instant (GCI) detection. Based on the algorithmic steps of DYPSA method and those set by the Multiscale-Product of LPC residual signal, MP(e(n)), method, it is proposed to compare the effect of Group Delay function followed by Dynamic Programming with that of the MP method, with a view to GCI detection. Results of this comparative approach are presented in terms of reliability, accuracy, and cost of execution in a clean environment. Also, the robustness of the performances of these two methods is compared in the presence of white and babble noise. According to all comparison made, we can conclude that the Group Delay function followed by Dynamic Programming is inefficient in the GCI detection compared to the MP method.