The major techniques for measuring jet noise have significant drawbacks, especially when including engine installation effects such as jet–flap interaction noise. Numerical methods including low order correlations and Reynolds-averaged Navier–Stokes (RANS) are known to be deficient for complex configurations and even simple jet flows. Using high fidelity numerical methods such as large eddy simulation (LES) allows conditions to be carefully controlled and quantified. LES methods are more practical and affordable than experimental campaigns. The potential to use LES methods to predict noise, identify noise risks, and thus modify designs before an engine or aircraft is built is a possibility in the near future. This is particularly true for applications at lower Reynolds numbers such as jet noise of business jets and jet-flap interaction noise for under-wing engine installations. Hence, we introduce our current approaches to predicting jet noise reliably and contrast the cost of RANS–numerical-LES (RANS–NLES) with traditional methods. Our own predictions and existing literature are used to provide a current guide, encompassing numerical aspects, meshing, and acoustics processing. Other approaches are also briefly considered. We also tackle the crucial issues of how codes can be validated and verified for acoustics and how LES-based methods can be introduced into industry. We consider that hybrid RANS–(N)LES is now of use to industry and contrast costs, indicating the clear advantages of eddy resolving methods.