Reproduce “Point-and-Play: Instrument Timbre Cloning from Music Mixtures” 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
Baotong Tian, Zhiyao Duan
Point-and-Play extracts a reusable timbre code directly from a musical mixture using an audio or text cue, then conditions a note synthesizer on that code and requested pitch and velocity. On 27 held-out NSynth instrument identities, text-cued generation reaches 0.706 CLAP cosine similarity to synthesis targets, above tested CLAPSep cloning at 0.560, but below an oracle-mask baseline at 0.726.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.