Sound hybridization techniques based on a deterministic plus stochastic decomposition model