Sound synthesis with mass-interaction physical modeling networks is known as a general paradigm capable of being the central part of complete software environments for both sound synthesis and musical creation. GENESIS 3, resting on the CORDIS-ANIMA formalism and developed by ACROE/ICA Laboratory, is the first environment of this kind. Using it, the artist may be facing an inherent problematic of every creation process: how to use a given tool in order to obtain an expected result. In our context, the question would be: “Considering a sound, which physical model could produce it?” Our work aims at formalizing this inverse problem and therefore at helping the user through his creative process. Therefore, we will consider how he could describe a sound (entry of the inverse problem), how to define a generator model based on mass-interaction physical networks and each one of its subcomponents (formal solution of the inverse problem), and, obviously, how to compute this solution considering an entry (resolution of the inverse problem). In this paper, we will develop each one of those three points and present the first algorithmic resolutions already implemented and used within GENESIS 3.