Our project develops the software OpenWInD for wind instrument making. A first feature is the prediction of the acoustical response of the instrument from the knowledge of its shape (bore and holes). This can be done in the harmonic (impedance computation) and temporal (sound computation) domains. It can account for various physical situations (non constant temperature, coupling with an embouchure, ...). Discretization is done in space with 1D spectral finite elements and in time with energy consistent finite differences. The second feature is the reconstruction of the shape of an instrument that fulfils a certain objective. This can be used for bore reconstruction, and instrument design. The latter is based on a strong interaction with makers and musicians, aiming at defining interesting design parameters and objective criteria, from their point of view. After a quantitative transcription of these criteria, under the form of a cost function and a design parameter space, we implement various gradient-based optimization techniques. More precisely, we exploit the fact that the sound waves inside the instruments are solution to acoustic equations in pipes, which gives us access to the Full Waveform Inversion technique (FWI) where the gradient is characterized as the solution to another wave equation. The computational framework is flexible (in terms of models, formulations, coupling terms, objective functions...) and offers the possibility to modify the criterion by the user. The goal is to proceed iteratively between the instrument makers and the numerical optimisation tool (OpenWInD) in order to achieve, finally, criteria that are representative for the makers. In the presentation, we will demonstrate and discuss some comparisons between measurements and simulation on real instruments.