Over the last few decades, radio detection has become one of the standard techniques to study high-energy cosmic-ray air showers. For the purpose of analysing the data, we heavily rely on Monte Carlo simulations. Upcoming dense radio array experiments such as LOFAR2.0 and SKA will, however, reach the limit of what is computationally feasible with these. Other techniques are available, based on macroscopic quantities, but their accuracy has thus far not been adequate to use them in precision analyses. In this contribution we present the latest update on the template synthesis approach, a hybrid model using both micro- and macroscopic inputs to synthesise the radio emission for an air shower with an arbitrary longitudinal profile. The method starts from the emission of a given shower and employs semi-analytical relations which only depend on the atmospheric depth at shower maximum and antenna position in order to transform it. Core to the template synthesis approach is the slicing of the atmosphere. By considering the radio emission from each slice separately, we only need to explicitly account for shower age effects. In previous work it was shown this could be done over a wide range of primary energy and across primary types for vertical air showers, with an accuracy of 10%. Here, we generalise the method to other zenith angles. We investigate the potential to synthesise between different geometries using a data set consisting of several hundreds of CORSIKA showers with primary energies ranging from $10^{17}$ eV to $10^{19}$ eV.