MuSE::A Fourier explanation of AI-music artifacts

​Research paper reading led by Jeff Gordon, MS Data Science: "A Fourier Explanation of AI-music Artifacts" by Darius Afchar, Gabriel Meseguer-Brocal, Kamil Akesbi and Romain Hennequin.

​Tools like Suno and Udio can generate a full song from a text prompt. Is there anything in the sound itself that shows a song was generated? Jeff walks through a paper that looks for the answer in how these models build audio and in the frequencies that end up in the result. No background needed. Reading and discussion, open to all NYU students.

​For those who aren’t NYU-affiliates or won’t be able to make it in person, please join us online:

​https://nyu.zoom.us/j/96543996147