Abstract

In this talk, examples of (nearly) 1/f behavior of sea-wave noise data and voiced speech residual signal will be presented. A wavelet-based method is used as a tool to analyze the power-law relationship. For speech residual signal, we particularly study some vowels ( i.e., /IY/, /IH/, /EI/, /EH/, /AE/, /ER/, /AH/, /AW/, /OA/, /OO/, /UW/ and /UH/) that are generated by several men and women. To extract the speech residual, we whiten the power spectrum of the speech signal by using a preemphasize filter and then perform the linear predictive analysis on the whitened speech to obtain the vocal tract parameters. The speech residual signal is obtained by the inverse filter. A wavelet decomposition technique is applied to the residual signal to obtain the wavelet coefficients. The power-law relationship is observed in the progression of the variances of these coefficients along scales. The self-similarity parameters (the slope of the progression) are then estimated. We investigate and compare the behavior of the self-similarity parameters for the speech samples of 40 men and 40 women. A speaker gender identification method is explored based on the self-similarity parameter. The random fractalness of the speech residual signals can be considered to be a key factor in their naturalness which may help us to explore designing more realistic synthetic signals for speech synthesis.

Biography

Tayfun Akgul is Professor of Electrical Engineering at Istanbul Technical University and is a freelance cartoonist. His cartoons are regularly published in IEEE Engineering in Medicine and Biology Magazine and IEEE Region 8 Newsletter and in the science edition of Turkey's the most prestigious newspaper Cumhuriyet. He currently teaches graduate and undergraduate level courses in Signals and Systems, Detection and Estimation Theory, Probability Theory and Stochastic Processes. He also conduct research in the area of signal/image processing.

His academic papers have appeared in Physical Review Letters, Signal Processing (EUROSIP), IEEE Transactions on Signal Processing, IEEE Transactions on Circuits and Systems, IEEE Transactions on Biomedical Engineering, Material Evaluation, and Materials Journals. After receiving his PhD in Electrical Engineering from the University of Pittsburgh in April 1994, Dr. Akgul became a NATO postdoctoral fellow at Pittsburgh. In 1997, he was Visiting Assistant Professor and later Research Associate Professor in the Electrical and Computer Engineering Department at Drexel University. In 1999, he returned to Turkey to take up the post of Chief Senior Researcher in the Information Technologies Research Institute at TUBITAK. Since 2002, he is with the Department of Electrical and Electronics Engineering at Istanbul Technical University.