Philippine E-Journals

Fractal Dimensions of Voice Patterns and Voice Recognition

Randy K. Salazar | Rosario L. Cabillada | Mark S. Borres | Anthony Jagures

Discipline: Computer Science, Engineering

 

Abstract:

The popularity and the convenience of using electronic communications have given rise to more transactions online. Despite the regular updates of safeguards, there are significant numbers of transactions that go awry. In the hotel business, forgeries and prank calls may be problematic, but there is nothing more distressing than to settle conflicts with guests. The lack of means to recognize, identify and verify callers exposes the transactions to pranks or to misunderstandings. In either case these frustrating transactions erode goodwill, which result in loss of future business. This study explores the use of fractal dimensions in characterizing the different facets of voice and speech dynamics. The different sinusoid samples intend to measure the physiological and the dynamics aspect of vocalization. Test results have shown that the differences of the group mean of the fractal dimensions of the voice wave patterns among the volunteers are significant. These also have shown the potential use of fractal dimensions in characterizing the voice patterns of different speakers and the eventual voice recognition or speaker identification.