Introducing a dataset of guitar amplifier sounds for nonlinear emulation benchmarking

摘要

Recent progresses made in the nonlinear system identification field have improved the ability to emulate nonlinear audio systems such as tube guitar amplifiers. As a straightforward comparison of different models cannot always be made, we propose a new reference dataset enabling to highlight the strengths and weaknesses of different nonlinear modeling methods. Our dataset gathers five different styles of guitar sounds passing through different guitar amplifiers with 10 steps of their gain parameter (i.e. the distortion level of the amplifier). Moreover, a Matlab function is also provided to obtain the desired input/output sounds in a matrix form.