Autotune is one of the most spread voice effect in the music industry, which consists in modifying, on the fly, the pitch of vocal (and instrumental) recordings. It was originally intended to correct out-of-key voices, but has been turned for music effect creation since then, the first example of such use being the single "Believe" by Cher in 1997. Since this first milestones in the music industry, the tools has been massively employed, faking the actual vocal performance of the singers.

The aim of this project is to develop an automatic tool that could be used to uncover the use of Autotune in any music excerpts. The project will first require a thorough characterization of the autotune effect, based on the analysis of specifically designed musical excerpts. Then, an automatic data processing algorithm will be developed and challenged on a set of calibration music files. Finally, the student will propose a methodology to reverse the effect of the autotune, with a view to, hopefully, unveil the real voice of singers.

Profile: Electrical Engineering, Computer Science

Prerequisite: Signal processing

Context: Theory (60%), data processing (40%)