We have developed high-fidelity CFD/CAA capability for the prediction of fan tone and broadband noise from an aero-engine. With a zonal URANS—wall-modeled LES approach, referred to as the Improved Delayed Detached Eddy Simulation (IDDES), the entire geometry of NASA's Source Diagnostic Test (SDT) fan rig including the nacelle is solved at approach, cut-back and take-off conditions. Multi-block grids with high-order structured finite-volume schemes are designed to resolve turbulence in the wakes and vortices as well as sound propagation in the bypass duct through the blade rows. By applying the Ffowcs-Williams and Hawkings technique with permeable surfaces, far-field sound-power spectra are computed, and by incorporating with the duct-mode-extraction technique, the radiated sound from the inlet and exhaust is decomposed into duct modes. These predicted fan-noise characteristics as well as aerodynamic performances are validated against NASA’s wind-tunnel test data. The fan-noise associated with the rotor-stator interaction is very well characterized: The tone power levels up to the second blade passing frequencies from both inlet and exhaust are predicted well across the engine speed, and the broadband power spectra are also predicted except for those from the inlet only at supersonic tip Mach numbers.