Nuance Labs debuts Emma, a full-duplex AI persona matching human reply timing 89% of the time
rohanpaul_ai · x · 2026-11-10
Nuance Labs launched Emma, an AI video persona whose reply timing matched real human speakers 89% of the time. She runs on Nuplex 1.0, a full-duplex audiovisual model that listens and speaks simultaneously, reacting facially while you're still talking.
Her face and voice come from a real actress who recorded how she actually listens, reacts and speaks over 18 months of research, giving Emma one coherent set of mannerisms: long blinks for emphasis, squints, drifting gaze, small emotional jumps, and a head that moves with the sentence.
Related event: Nuance Launches Full-Duplex Model Nuplex 1.0 and AI Human Emma(7 posts)→
More from Multimodal
- 3D celebrity caricature prompt shared: exaggerated head, recognizable face — azed_ai · 2026-11-10
- Agent given a vocal-synth repo builds branching acoustic-tube singing synthesis — TinfoilTricorn · 2026-11-10
- Sillage 3D: open-source browser engine renders Gaussian splats as brush strokes — janusch_patas · 2026-11-10
- Testing Fable 5.2 image model: creator shows off results made for 1 token — alex_verem · 2026-11-10
- One reference image, one prompt: MiniMax H3 Image rebuilds a Porsche poster as an F1 design — JaynitMakwana · 2026-11-10
- 29-minute claymation film built in 7 days with Opus 5.5 Max on a GTX 1660 Ti — Humble-Scale4941 · 2026-11-10