The article introduces a new method for detecting and quantifying heterogeneous mixtures of dynamic processes within phylogenetic trees, which can identify multiple time-varying diversification processes without specifying their locations in advance. The method uses reversible-jump Markov Chain Monte Carlo (rMCMC) to explore model subspaces with varying numbers of distinct diversification regimes. It assumes that evolutionary regime changes occur across phylogenetic branches under a compound Poisson process, accounting for rate variation through time and among lineages. The method is demonstrated on simulated datasets and compared to the MEDUSA model, showing superior performance in inferring complex mixtures of time-dependent, diversity-dependent, and constant-rate diversification processes. An empirical example analyzes the speciation and extinction dynamics of modern whales, highlighting the method's utility in exploring macroevolutionary dynamics across large phylogenetic trees.The article introduces a new method for detecting and quantifying heterogeneous mixtures of dynamic processes within phylogenetic trees, which can identify multiple time-varying diversification processes without specifying their locations in advance. The method uses reversible-jump Markov Chain Monte Carlo (rMCMC) to explore model subspaces with varying numbers of distinct diversification regimes. It assumes that evolutionary regime changes occur across phylogenetic branches under a compound Poisson process, accounting for rate variation through time and among lineages. The method is demonstrated on simulated datasets and compared to the MEDUSA model, showing superior performance in inferring complex mixtures of time-dependent, diversity-dependent, and constant-rate diversification processes. An empirical example analyzes the speciation and extinction dynamics of modern whales, highlighting the method's utility in exploring macroevolutionary dynamics across large phylogenetic trees.