Reproduce “UniAVC: Unified Audio-Video Creation via Progressive Training and Asymmetric Scheduling” 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
Kien T. Pham, Yuhao Liu, Chong Hou Choi, Zhenqi He, Yanghao Wang, Chenyang Lei, Rui Liu, Qifeng Chen and 1 more
UniAVC combines audio and video generation and editing. It trains branches separately before using paired data for cross-modal links, and maps one shared progress counter to different audio and video denoising schedules. On JAVEditBench, default shifts (2, 9) score higher than equal shifts (1, 1) on AQ (2.59 versus 2.48) and AVQ (3.85 versus 3.67), while visual quality is similar.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.