In addition to the sparsity property of a wireless channel, the channel multipath components are modeled as clusters of multirays due to scatterers distributed in the environment. This is a typical channel configuration that future multiple-input multiple-output (MIMO) wireless communication systems will have to cope with. Hence, there is a strong need to design new appropriate channel estimation techniques. A cluster of multirays can be characterized by its mean delay and its delay spreading. In a previous work, we have focused on estimating the mean delays of the different clusters based on a deterministic channel model. For the work to be complete, we propose in this paper to estimate the standard deviation of the delay spreading of each cluster based on a stochastic model, exploiting time delays distribution of the scattered signals.