ABSTRACT
Generalization of the signal representation in terms of sinusoidal functions leads to amplitude and frequency modulated signal model, where each sinusoidal component can have time-varying amplitude and frequency in the representation. The combination of complex amplitude modulated and frequency modulated signal models has been applied in order to represent real speech signal in parametric manner. These complex amplitude modulated and frequency modulated signal models have been proposed for online processing of speech signals. Time-frequency distribution represents the energy density of the signal simultaneously in time and frequency-domains, which can be used in the similar way like any other density functions.
