The paper is dealing with the prediction of fan broadband self-noise. Two mechanisms are investigated, namely trailing-edge noise resulting from the scattering of blade boundary-layer turbulence at the trailing-edge, and vortex-shedding noise associated with the von Kármán street formed in the near wake due to bluntness. The emphasis is on analytical models based on the acoustic analogy. The models are understood as post-processing tools of unsteady flow features determined first, according to the general scope of hybrid methods in aeroacoustics. The statistics of the far-field acoustic pressure is expressed as a function of either a wallpressure statistics or a near-wake velocity statistics, needed as input data. The data can be obtained from dedicated experiments for validation purposes, or from flow computations for pure prediction studies. An industrial airfoil and a thick plate at different angles of attack are chosen as test cases, for which both experimental results and computations performed elsewhere are available. The analytical noise predictions agree well with the measurements. They are compared to a pure numerical approach and their limitations, as well as their advantages for fast-running low-noise design in an industrial context, are discussed.