In recent years, the Barkhausen effect has been indicated as a promising tool to investigate and verify the ideas about the self-organization of physical complex systems displaying power law distributions and 1/f noise. When measured at low magnetization rates, the Barkhausen signal displays 1/fα-type spectra (with α=1.5÷2) and power law distributions of duration and size of the Barkhausen jumps. These experimental data are quite well described by the model of Alessandro et al. which is based on a stochastic description of the domain wall dynamics over a pinning field with brownian properties. Yet, this model always predicts a 1/f 2 spectrum, and, at the moment, it is not clear if it can take into account possible effects of self-organization of the magnetization process. In order to improve the power of the model and clarify this problem, we have performed a thorough investigation of the noise spectra and the amplitude distributions of a wide set of FeCoB amorphous alloys. The stationary amplitude distribution of the signal is very well fitted by the gamma distribution P(ν)=νc−1 exp(−ν)/Γ(c), where ν is proportional to the domain wall velocity, and c is a dimensionless parameter. As predicted in Ref. , this parameter is found to have a parabolic dependence on the magnetization rate. In particular, the linear coefficient is related to the amplitude of the fluctuations of the pinning field, a parameter which can be measured directly from the power spectra. In all measured cases, the power spectra show α exponents less than 2, and thus poorly fitted by the model. Actually, the absolute value of the high frequency spectral density is not consistent with the c parameter determined from the amplitude distribution data. This discrepancy requires to introduce effects not taken into account in the model, as the propagation of the jumps along the domain wall. This highly enhances the fit of the data and indicates effects of propagation on the scale of a few millimeters. These results are analyzed in terms of new descriptions of the statistical properties of the pinning field based on fractional brownian processes.