As in many industries, NVH reduction in the automotive market requires car manufacturers to adapt their designs and development methods. In particular duct system design necessitates accurate methods to capture the resonance frequencies and address noise phenomena. To do so, engineers evaluate the quality of the components through numerical simulations and validate against real-life measurements. Two main challenges arise: the simulations are often ran by non-numerical experts, who might have difficulties with setting up the models. Secondly, the computation techniques, together with the model preparation, need to be efficient to improve productivity. In this paper, an automated workflow is presented to democratize numerical techniques in duct design and sound transmission loss computation. It gives the possibility for non-experts to easily set up and solve their models with few input parameters. Furthermore, effective numerical techniques based on high-order shape functions with a-priori error estimator are used to optimize performance and accuracy: FEMAO and BEMAO (Finite/Boundary Element Methods with Adaptive Order). Two models are exposed: a simple duct line with three Helmholtz resonators, and a more complex intake system. The numerical results of the different methods are compared, as well as measurements.