The processes driving continental crust formation are not yet fully understood. One of the fundamental keys necessary to investigate the enigma is represented by crustal composition. The Vp/Vs ratio from seismic receiver functions or tomography studies is a powerful tool to constrain the crustal composition. However, to date only qualitative relationships between Vp/Vs and composition have been proposed. We present a quantitative interpretation of the Vp/Vs in terms of major oxide components, based on thermo-elastic constrained modelling of rock phase equilibria and physical properties. The geophysical-petrological approach is implemented in the new release of the software package LitMod, which now allows for integrated and self-consistent modeling of the entire lithosphere (crust + lithospheric mantle) and upper mantle. Forward modelling of the Vp/Vs, based on petrology and thermodynamics, reveals that, as expected, mafic compositions have higher Vp/Vs than felsic ones. However, in high temperature settings (surface heat flow > 75 mW/m2), the quartz alpha / quartz beta transition strongly increases Vp, leaving Vs almost unaltered, leading to SiO2-rich compositions displaying Vp/Vs values higher than those associated with mafic compositions. Additionally, we highlight the importance of H2O, the presence of which stabilizes amphibole (in place of pyroxene), characterized by a relatively low Vp/Vs. If H2O is present, mafic compositions show Vp/Vs ratios that are comparable to those produced by anhydrous SiO2-rich compositions. The destabilization of amphibole (in favour of pyroxene) generates a sharp seismic discontinuity, potentially detectable by, for example, seismic refraction and receiver function investigations. We invert the Vp/Vs ratio for composition and hydrous state of the crust in the Southern African cratons. Our results show that the Kaapvaal craton, Archean in age, has an intermediate (SiO2 60 wt%) composition. The finding has implications on our understanding of the mechanisms responsible for crustal formation as an intermediate bulk composition is potentially generated by subduction-related processes. We conclude that, already before the Meso-Archean (3.2 Ga), subduction related crust-forming processes were active in the area and potentially also in other cratons.