The authors connected a pretrained NVIDIA speech encoder to a DiffusionGemma text model.
Diffusion separates response timing from content in full-duplex speech
Diffuplex uses an autoregressive policy to manage conversational timing while a discrete diffusion model generates response tokens in parallel.
Big Tech
Chen Chen · Pin-Jui Ku · Ankita Pasad · Ante Jukić · Zhehuai Chen · Boris Ginsburg
NVIDIA
Research Digest··2 min read
Chen et al.
Why this paper
From NVIDIA
In one line
Diffuplex decouples when to speak from what to speak, enabling fast and intelligent full-duplex spoken dialogue.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§