F-Actor is the first open instruction-following full-duplex conversational speech model, trained by freezing the audio encoder and finetuning only the language model on 2000 hours of data, with public release of model and code.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
F-Actor: Controllable Conversational Behaviour in Full-Duplex Models
F-Actor is the first open instruction-following full-duplex conversational speech model, trained by freezing the audio encoder and finetuning only the language model on 2000 hours of data, with public release of model and code.