Large scale biodiversity patterns result from the historical processes of speciation and extinction. In particular, the balance between speciation and extinction rates determines how species richness varies through time, across species groups, and geographical regions. Phylogenetic diversification analyses, which rely on fitting stochastic birth-death processes to phylogenetic data, can be used to estimate these macroevolutionary rates from the phylogenies of extant species, potentially further informed by paleodata. I will present recent developments that model fine-scale variations in speciation rates and that can combine neontological and paleontological evidence. Applied to empirical data, these models reveal a wide variation in speciation rates across lineages. While several models have been developed to explain these variations by differences in specific traits or abiotic and biotic conditions, models that would help us better understand the actual processes that control diversification rates are lagging behind. Speciation research at the microscale has focused on understanding the establishment of reproductive barriers, but there is increasing evidence that variations in macroevolutionary speciation rates are poorly explained by variations in the rate at which populations acquire reproductive isolation. I will present recent developments that aim at understanding i) how variations in the rates at which reproductive isolation is initiated, at which populations acquire reproductive isolation, and at which incipient species go extinct combine to explain macroevolutionary speciation rate, and ii) how population sizes, mutation rates and the mode of speciation impact this latter relationship.[-]