Multimodal Art Projection (m-a-p)
Open research collective publishing music and multimodal models on Hugging Face (1.35k followers at read time): the MERT music encoders, YuE and YuE2 (song generation), and sheetsage2, the transcription model that brings them into this spoke. A third source now reaches this org secondhand: yue2-studio, a wrapper app that runs YuE2 and also puts sheetsage2 to work for covers — the first page here to show a m-a-p model actually being used rather than just described.