Novel chaotic titration method for analysis of normal and pathological voices

This paper proposes a novel approach to quantify the amount of chaos that is present in the speech signal to analyze the normal voice from that of the pathological voice (primarily, the voiced regions). The use of chaotic titration as a method to estimate the chaotic nature of the speech signal is demonstrated. The noise limit (NL) value denoted as the “noise ceiling” of the signal, corresponds to the amount of chaos in the system that produced the signal (e.g. speech production mechanism). Experimental results confirm that NL is more in case of pathological voice as compared to normal voice. Poon and Barahona found this method to perform better than the other method such as Lyapunov exponent for quantification of chaos. In addition, the proposed method requires use of less number of sample values for analysis; which reduces computation complexity.