We have shown previously that the statistics of the radio-frequency (RF) signals may be faithfully modeled through the so-called K(RF) distribution, in situations ranging from fully to partially-developed speckle. We demonstrate in this paper that the generalized Gaussian provides a reliable and computationally convenient approximation of the K(RF). The performance of the parameters estimators for the two distributions is evaluated and compared in terms of their bias and variance through numerical simulations. This framework is applied to the modeling of echocardiographic data. The ability of the generalized Gaussian to model RF signals from cardiac tissues (myocardium) and blood regions is demonstrated on data acquired in vivo.