Recently M. Roger has studied a Katana blade. The latter is the name of the traditional single-edge, slightly curved sword used in Japanese martial arts. When the Katana is handled for cutting, the blade is moved through the air at a relatively high speed and a hiss or hushing sound can be heard. This sound depends on the angle of attack, the flow speed and the blade-profile design. Furthermore, different configurations of the back-edge exist, with or without grooves. The bluntness is responsible for the acoustic signature referred as the vortex-shedding and the grooves are adding high-frequency sound (cavity-tones). M. Roger made acoustic measurements on several Katana configurations with a wind tunnel jet velocity varying from 12 to 35 m/s yielding both low Reynolds and Mach numbers. The present work is a first attempt at deciphering the various noise mechanisms at hand observed in the above measurements achieved by M. Roger. It focuses on the comparison between two configurations of the Katana blade at a single incidence of 0°: a case with two symmetric grooves and another one with a single groove. In a first step, we used Fluent 6.3 with a k-ω SST turbulence model to make a complete modeling of the small ECL anechoic wind tunnel in a RANS simulation. Then, a 3D mesh has been generated by an extrusion normal to the Katana profile over a spanwise extent of 10% of chord length and a full 3-D LES is performed on the blade, using the boundary conditions extracted from the RANS calculation. The LES on the symmetric two-groove configuration is already showing well-developed turbulent structures when looking at the iso-contours of the Q factor. The flow structure already suggests that this configuration should provide a proper test case for the analytical vortex-shedding noise model proposed by Roger et al.