Exciting invited talk today by Megan Wei from Brown University on her #ISMIR2024 work, 'Do Music Generation Models Encode Music Theory?'! Their novel SynTheory dataset & probing framework reveals how models like Jukebox & MusicGen encode fundamental music concepts.

Hyped that 3/3 papers w/the folks @ucsd-musaic.bsky.social are accepted at #ICASSP2025! PDMX: Public Domain Symbolic Music arxiv.org/abs/2409.10831 CoLLAP: Long-Context CLAP (~5 min) arxiv.org/abs/2410.02271 FUTGA-MIR: long music understanding for MIR tasks (arxiv soon) Next stop, India!🇮🇳

We just created a Bluesky starter pack featuring people and groups working at the intersection of AI and music, covering both symbolic and audio approaches. Let us know if you'd like to be added or removed! go.bsky.app/PBvFCxa

We’ve just launched our official Bluesky account! We’ll be sharing exciting AI x Music projects from UCSD (PIs: Julian McAuley, Taylor Berg-Kirkpatrick, Shlomo Dubnov).