Numerical design of a trumpet lead pipe: from the numerical model to the evaluation of the hardware prototype

In a previous work, a machine learning model was trained using numerical simulations in order to predict intonation and playability descriptors of trumpets from the bore geometry [1].Using this technology, a numerical design procedure was conducted where a new leadpipe geometry was proposed using a bi-objective optimization on the Equivalent Fundamental Pitch (EFP) and minimum blowing pressure (Pth), computed over five regimes.This recommendation led to a leadpipe numerical prototype that was manufactured by Yamaha Corporation, Japan.In this paper, we present recent investigations conducted on this hardware leadpipe prototype in order to compare the performances of the real instrument with the numerical expectations in the design phase.At first, impedance measurements of the prototype were performed, and sound simulations were computed from these measurements to assess the EFP and Pth descriptors.They were compared to the results of the numerical prototype.In a second step, playing tests with musicians were conducted in order to assess perceived differences between the original and prototyped leadpipe.These results are discussed in light of the potential sources of uncertainties through the whole procedure, from the model to the perception.