Reproduce “Getting Motif-ated: Controllable AI Compositions from Injected Motif Prompts” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chao Peter Yang, Cynthia Rudin, Yue Jiang, Simon Mak, Stephen Ni-Hahn
MotiGen lets composers specify an abstract melodic motif and incorporates it into monophonic music by adding attention bias and fine-tuning a pretrained model through focused crops and full scores. In an automatic test across 24 motifs, the full recipe contained the requested pattern in 92.3% of generations, versus 77.8% without attention bias. This measures motif inclusion, not musical quality.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.