The historical detection of a gravitational wave signal from a double neutron star merger a couple of years ago ushered in a new era for the study of these fascinating compact objects. With more double neutron star detections expected in the incoming years, the impact of binarity on stellar evolution is now a key-element to trace back the evolution of these highly magnetized spinning tops. In high mass X-ray binaries, the transient albeit decisive phase preceding double neutron star systems, a neutron star orbits a supergiant O/B star and captures part of its stellar wind. We aim at characterizing the mass and angular momentum transfer to the accreting neutron star. Our numerical simulations of wind accretion onto the neutron star reveal a complex 3D geometry : depending on the efficiency of the wind launching, on the extent of the neutron star magnetosphere and on the cooling mechanism of the flow, a wind-captured disc can form around the accretor, with dramatic consequences on the spinning up/down of the neutron star Because the spin of the neutron star retroactively controls accretion and outflows through magneto-centrifugal gating effects, a better understanding of the angular momentum loss and transfer in wind-fed high mass X-ray binaries is required to interpret the observed torques and evaluate their capacity to eventually lead to a neutron star merger within a Hubble time. In this talk, I will discuss the conditions of formation of a wind-captured disc and the impact of the inhomogeneities in the wind (or “clumps”) on the time variability of the X-ray flux in high mass X-ray binaries. I will also present how slow winds can provoke a significant enhancement of the mass transfer rate to the accretor, up to levels suitable for the super-Eddington accretion regime observed in ultra-luminous X-ray sources.