Marcel Vélez Vásquez is a PhD candidate at the University of Amsterdam, specializing in explainable artificial intelligence for playability from musical audio. In this episode, he explains all about what AI for music is, and how it is different from textual AIs like ChatGPT. Artificial intelligence can be of help at many facets in the world of music: write and master songs, mix instruments together, but also predict whether an audio fragment could potentially grow out to be a smash hit, or help a beginner choose the next song to learn on their instrument. We also talk about the more challenging sides of his field, like using someone else's voice and the Western bias in available data.

Tracklist, can you guess which are easy or hard to play on the guitar?:
1. Stevie Wonder - Do I Do
2. Jimmy and David Ruffin - Stand By Me
3. X Ambassadors, Elle King, Wiz Khalifa - Not Easy
4. Can AI Kick It? - Abbus (with AI generated music video: https://www.youtube.com/watch?v=YbsWkhcJjg4)

Artificial Intelligence for Music @Echobox Radio

Talk That Science is an Amsterdam-based science show by a group of science and music enthusiasts.  We play the nicest tunes while talking about the coolest research!

Live on echobox.radio every month