Reproduce “F-Actor: Controllable Conversational Behaviour in Full-Duplex Models” 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
Maike Züfle, Ondrej Klejch, Nicholas Sanders, Jan Niehues, Alexandra Birch, Tsz Kin Lam
F-Actor makes a spoken assistant’s voice, topic, turn-taking behavior, and choice of who starts explicit prompt inputs for full-duplex conversation. It processes user and system audio streams and can follow start instructions reliably, reaching 99.8% correct starts in its best reported setup. Requested backchannel and interruption counts remain approximate, with correlations of 0.54 and 0.25.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.