Andreas Spanias

TL;DR: It is shown that 10 out of 15 seizures in one patient with temporal lobe epilepsy were predictable with an average predictability time of about 36 minutes, derived from the created KLT dimensionality profiles.

TL;DR: This book describes several modules of the Code Excited Linear Prediction (CELP) algorithm using the Federal Standard-1016 CELP MATLAB(r) software to describe in detail several functions and parameter computations associated with analysis-by-synthesis linear prediction.

TL;DR: Two fundamental enhancements in a hybrid audio signal model consisting of sinusoidal, transient, and noise (STN) components are presented, including a novel application of a perceptual metric for optimal time segmentation for the analysis of transients and a new methodology for ranking and selection of the most perceptually relevant sinusoids.

TL;DR: A collection of software educational tools for introducing speech and audio compression techniques to undergraduate and graduate students that enable online simulations of speech compression algorithms that are being used in digital cellular phones, Internet streaming applications, teleconferencing, and voice over Internet protocol (VoIP) applications are presented.

TL;DR: It is shown that among the noise distributions considered, the estimators are asymptotically efficient only when the noise distribution is Gaussian, and Simulation results corroborate analytical results.