Detecting the longest periodic timescale in normal vowels

Abstract

Normal vowels are confirmed to have irregularities, where various methods have demonstrated that there is chaotic property hidden in them. However, little attention has been given to such phenomenon that, in short timescale, the vowel single usually shows periodic character and when the time length is long enough, it converts to chaotic dynamics. In this paper, we aim to search the longest time window that the normal vowel keeps its periodic dynamics. We employ a novel pseudoperiodic surrogate algorithm to test whether the normal vowel segment is consistent with the periodic orbit. The results reveal that the longest time window is invariant for different vowel data, and for most vowel data segment, they keep periodic dynamics given that their timescale is between 25ms and 35ms by investigating typical Chinese normal vowels from male and female subjects.