Reproduce “AVCC: Audio-Visual Conversation Corpus for Turn-Taking Prediction with a Human Benchmark” 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
Haotian Qi, Gabriel Skantze
The authors extend AVCC into a benchmark for turn prediction in two- and three-speaker conversations recorded with one camera and one microphone. The benchmark adds a speaker-disjoint test set and human judgments at the same mutual-silence gaps used to evaluate models. On hold/shift, MuVAP scores 67.9%, versus 66.5% for RoleVAP; these results do not isolate video’s contribution or establish live-robot performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.