This paper presents the depth inversion of Rayleigh wave group velocity to obtain an S-wave velocity model from seismic ambient noise cross-correlation in Makran subduction zone. The Rayleigh wave measurements for the period range 5-50 s were extracted from two seismic data sets: ambient noise cross-correlations and regional earthquakes recorded on 65 broadband stations. In this study, the Bayesian Monte-Carlo algorithm was applied to invert the Rayleigh wave group velocities into 1D shear-wave velocity (Vs) profiles, Juxtaposing the all 1D models generated, we composed a 3D velocity model for the study region which is the first high resolution crustal image beneath the Makran subduction zone. The low velocity anomalies of the upper crust correspond to regions of thick sediments. Results show that the deepest sedimentary basin is in the Makran accretionary Wedge. Also, the high velocity anomalies are observed at the depth of approximately 30 km beneath Gulf of Oman and Makran Accretionary Wedge, that correspond to high velocity oceanic mantle.