The constant evolution in the automotive sector to achieve more eco-friendly vehicles has induced the development of more efficient systems with new components and innovative materials. To evaluate the impact of these technologies or to improve them in terms of NVH performances, acoustic engineers rely on experimental tests and numerical computations. In this context, the use of experimental noise sources identification and characterization methods can provide interesting approaches. However, classical methods usually used in industry like the Nearfield Acoustical Holography (NAH) or the Beamforming techniques are quickly limited, in particular in terms of precision in localization, for such analysis support. The presented method, named M-iPTF for Mixed inverse Patch Transfer Functions, is more suitable as it is able to localize and quantify all acoustic source fields directly on the real geometry of a complex structure. As a result, it offers a more accurate noise sources identification, the possibility of ranking sources by the computation of the radiated power by parts, and a more efficient coupling between experiments and simulations to upgrade a model or to reuse experimental data in a larger virtual model. The M-iPTF principle is based on an inverse acoustic problem, formulated from the application of the Green’s identity on a closed virtual volume defined around the source. It only needs simple pressure measurements, which can be performed in an uncontrolled environment, coupled to a numerical modelling. This article will briefly present the theoretical background of the method, before illustrating its benefits as an analysis support technique in an industrial application: a reduced engine block excited by an electrodynamic shaker.