Reproduce “A Study on Improving Multi-class Audio Source Separation Via Decoupled CLAP Query Optimization and an Automated Data Engine” 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
Amirhossein Hajavi, Hanhee Lee, Pushya Jain, Sky Qiao, Emmanuel Ko, Yuanhao Yu, Irina Kezele
The authors present a language-queried audio source separation system. They screen candidate training examples with caption-based checks and optimize CLAP class-control embeddings. In seven-class evaluation, mean SDRi rises from 2.753 for the baseline to 5.545 after refinement and 5.610 after control optimization. The added gain is small, and optimized controls do not improve every class.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.