Characterization of healthy and pathological voice through measures based on nonlinear dynamics
IEEE Transactions on Audio, Speech, and Language Processing
Automatic detection of hypernasality in children
IWINAC'11 Proceedings of the 4th international conference on Interplay between natural and artificial computation: new challenges on bioinspired applications - Volume Part II
Hi-index | 0.00 |
A novel way for characterizing hypernasal voices by means of nonlinear dynamics is presented considering some complexity measures that are mainly based on the analysis of the embedding space. After characterization, feature selection is performed using two strategies, Principal Components Analysis (PCA) and Secuential Floating Feature Selection (SFFS); classification between healthy and hypernasal voices is carried out with a Soft Margin - Support Vector Machine (SM-SVM). The database used in the study is composed of the five Spanish vowels uttered by 266 children, 110 healthy and 156 labeled as hypernasal by a phoniatrics expert. The experimental results are presented in terms of accuracy, sensitivity and specificity to show in a quantitatively manner, how stable and reliable is the methodology. ROC curves are also included to present a widely accepted statistic for the accuracy of the system. According to the results, nonlinear dynamic theory is able to detect hypernasal voices, and would be worth to continue developing this kind of studies oriented to automatic detection of pathological voices.