Control of a circular cylinder flow by rear-attached solid/perforated splitters has been experimentally investigated using Particle Image Velocimetry (PIV) and far-field microphones, respectively. The Reynolds number is Re=2.7x10(4)-6.8x10(4) based on the cylinder diameter D, and the effects of splitter length L (L/D=0-4.0), porosity sigma (0%-22%), and deflection angle alpha (0 degrees-30 degrees) on noise and flow characteristics are reported. The literature seriously lacks the combination effects of these parameters on bluff-body wakes, especially the flow mechanism on noise variations. Acoustic results show that for a solid splitter, the optimal noise reduction of 18dB is achieved when L/D=1.0-1.5, whereas when 2.0 <= L/D <= 4.0, the noise increases rather than decreases, consistent with the literature. At L/D=1.0, deflecting and/or perforating the splitter would not further reduce the noise and, on the contrary, weaken the noise control efficiency. However, at L/D=2.0, a rear-half perforation with sigma >= 18.2% or a deflection angle of alpha >= 30 degrees individually reduces noise by more than 10dB compared to the bare cylinder. Combining both methods at L/D=2.0 eliminates vortex-shedding noise when sigma >= 11.6% and alpha >= 20 degrees. PIV measurements reveal that noise reduction at L/D=1.0 results from reduced vortex strength in the flow, while noise increase at L/D=2.0 is attributed to vortex-splitter trailing edge interaction, which is mitigated by the perforation and/or the deflection of the splitter. The study suggests that, in general, appropriately perforating and deflecting longer splitter plates (L/D >= 2.0) effectively reduce bluff-body noise, while solid, shorter splitter plates (0.5 <= L/D <= 1.5) positioned streamwise achieve significant noise reduction. These insights offer valuable noise control strategies for various engineering applications.