The other contributors to this discussion have a point but noise source contribution cannot be accurately quantified by simply placing a microphone close to the noise source. For one the closer you measure to the source you are measuring in the near field and secondly you cannot eliminate the contribution from other sources. Decibels (dB) is a non linear scale, i.e. 60dB plus 63dB does not equal 123dB. As a general rule any measurement made that is more than 10db less than the max measurement will have little or no effect to the overall noise level recorded, i.e. 50dB plus 60dB equals approximately 60dB. Also two measurements made with the same overall level of say 60dB can sound completely different depending on the spectral content. There are many excellent and reliable test and simulation software packages but regardless of what equipment you use unless you take good accurate repeatable measurements to obtain good raw data the software you use won't make any difference. The rule of gold in, gold out applies. You also have to consider the noise path, is it structure or airborne or both. Once the noise paths and sources are quantified then you can examine the spectral content to identify problem frequencies. Engine noise in a 4 cylinder engine will typically exhibit booming noise generated by the out of balance forces and is dominated by the second order. However there can be numerous other sources such as intake resonances, exhaust noise, shell noise, heterodyning noise caused by crankshaft bending or as in earlier times by unequal manifold ducts. Then there are ancillary noises such as pumps, alternator etc. In general changing the noise source can be difficult and expensive, palliative treatments to reduce the noise path maybe more practical, e.g. if the problem is low to mid frequency then applying heavy layer insulation to the cabin and damping to lightly damped less stiff panels will help. If on the other hand the problem is high frequency, typically above 1KHz then applying absorption materials will be the route to go. Car engines are much more refined than forklift engines in general, employing active mounts and advanced algorithms however some manufactures are better than others also some engines are direct transplants from the automotive sector. Ancillary noises can be tested by removing components and testing to see the effect on the noise level. Where parts cannot be removed it may be possible to apply noise treatment to them, measure the noise and quantify their contribution. It may be possible to make some prototype parts e.g. intake branch or helmholtz resonator to quantify intake noise. If you have access to CAD data and the financial resources then with the help of a consultancy such as Ricardo, AVL, Pelzer, Ove Arup, Rieter or some other group a model could be developed to simulate noise levels and carry out what-if scenarios. Although this is expensive, time consuming and only applicable if you were the manufacture and wanted to refine existing and future models.