Jet noise remains a key target for aircraft noise reduction in the foreseeable future. While being extremely challenging, the requirement of predicting both low and high frequency noise spectra is increasingly important for design purposes. Novel approaches are needed to overcome the current numerical limitations in capturing the required broad noise spectra. Once su ciently resolved, the energy contents of the numerically simulated near-and far-field sound pressure have intrinsic correlations among di↵erent levels of grid resolution. The present work explores the potential of such correlations to broaden the spectral prediction. The noise radiated from high subsonic turbulent jets is investigated using large-eddy simulation. The 3-D filtered compressible Navier-Stokes solutions are obtained for an axisymmetric and a serrated nozzle on successively refined multi-resolution grids, ranging from 5 to 80 million grid points. The radiated far-field sound is computed using the Ffowcs Williams -Hawkings (FW-H) surface integral method. Fourier decomposition for pressure near-field is applied to help identify the location of the sound source regions and the dominant directions of propagation, which provides a more thorough understanding of the e↵ect of the grid resolution on the numerical cut-o↵ frequencies of the far-field spectra. Further analysis of the far-field spectra and of their azimuthal modes confirms that a novel strategy to obtain a broadened overall sound spectrum is possible, at reduced computational cost, from a combination of multiple spectra from successively refined grids.