How to predict the sound that a piece of wood will make when transformed into a violin board? What shape should you give it to make it sound best? Artificial intelligence takes care of it. This is the conclusion reached by the researchers of the Musical Acou-stics Lab of the Politecnico di Milano, who have just published their study in the journal Nature Scientific Reports. In the article “A Data-Driven Approach to Violinmaking” the Chilean physicist and luthier Sebastian Gonzalez (postdoc) and the professional mandolinist Davide Salvi (PhD student) show how a very simple neural network is able to predict the vibratory behavior of violin tables starting from a limited number of geometric and mechanical parameters of the table itself. “The ability to predict how a violin with a certain shape sounds can really be a game changer for luthiers, as it will not only help them do better than the ‘grand masters’, but will also help them explore the potential of new designs and materials “says Sebastian Gonzalez. “This research – he adds – has allowed us to take the first steps on this path, showing how artificial intelligence, physical simulation and craftsmanship can come together to shed light on the art of violin making”. Researchers recall that violins are extremely complex objects, and their geometry is defined by the external shape and curvature along the vertical and horizontal directions of the plank and bottom. The inspiration for this study came from a historical drawing that is part of the collection of the Museo del Violino of Cremona. In particular, a model has been developed that draws the external form as the conjunction of arcs of nine circles. Thanks to this, and to an efficient model of curvature based on Stradivari’s Messiah violin, the researchers were able to design a violin table with only 35 parameters. By randomly varying these parameters, such as radius, position of the center of the circumferences, curvature, thickness and mechanical characteristics of the wood, a dataset of violins has been constructed on the computer, which includes historical violin forms as well as forms never used in violin making. These forms of tables constitute the input of the neural network. The term neural networks refers to a series of algorithms that mimic the behavior of a human brain and aim to identify relationships within large data resources. They are used in many applications. In their simplest form, they are a collection of nodes (neurons) in a network that processes input data and output results, and the way they do this is to vary the numerical parameters that define their behavior. , advanced vibratory modeling tools were then used to determine the acoustic behavior of each violin in the dataset (a set of data used to train the neural network, consisting of input and output data that the neural network must produce). The next step was to understand if a simple neural network is able to predict acoustic behavior starting from the parameters of the table. The response was positive, well beyond expectations with an accuracy close to 98%. The work therefore offers a promising instrument in the hands of the luthiers of Cremona and of the whole world. Using the neural network allows us to predict how a given piece of wood would “sound” if transformed into a board with a certain shape and could also be used to design two violins of different wood so that they sound the same. This will make it possible in the future to select the best wood for a particular violin model, an operation which today is performed on the basis of purely aesthetic considerations. The project was financed by the Cultural District of the Violin Making of Cremona.