Reproduce “PianoKontext: Expressive Performance Rendering from Deadpan Context” 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
Dmitrii Gavrilev
PianoKontext renders note-constrained piano scores as expressive performances at selected lengths. It encodes deadpan score audio and human performances into compressed latents, then uses precomputed dynamic time warping alignments to train variable-length generation without requiring equal durations. The method still depends on paired data. On held-out piano tests, it beats CFG Bridge but remains below human reference scores.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.