AI Paper Atlas

Research map / 3D vision / Face and shape

Talking Together: Synthesizing Co-Located 3D Conversations from Audio

Mengyi Shan, Shouchieh Chang, Ziqian Bai, Shichen Liu, Yinda Zhang, Luchuan Song, Rohit Pandey, Sean Fanello, Zeng Huang

CVPR 2026 · 3D vision › Face and shape · approach: Speech-driven 3D facial animation · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add second-stage fine-tuning on high-quality lip data (ablation)dual-stream diffusion model · Dyadic test set, ablationFD: 60.12 -> 21.71Improved
add second-stage fine-tuning on high-quality lip data (ablation)dual-stream diffusion model · Dyadic test set, ablationLip-Sync preference (%): 17.2 -> 78.3Improved
add conversation data pre-training on top of single-person data (ablation)dual-stream diffusion model · Dyadic test set, ablationFD: 50.45 -> 21.71Improved
add speaker role embedding (ablation)dual-stream diffusion model · Dyadic test set, ablationFD: 35.92 -> 21.71Improved
add inter-speaker cross-attention (ablation)dual-stream diffusion model · Dyadic test set, ablationP-FD: 40.87 -> 22.56Improved
add auxiliary eye gaze loss on high head-rotation samples (ablation)dual-stream diffusion model · Dyadic test set, ablationP-FD: 42.90 -> 22.56Improved
add auxiliary eye gaze loss (ablation)dual-stream diffusion model · Dyadic test set, ablationTRAN MSE: 2.59 -> 1.50Improved
train on dyadic conversation data instead of single-person modeldual-stream model vs Ours (Single) · Dyadic test set (Table 2)vMSE (FULL): 2.23 -> 2.28Mixed
use full interactive model instead of best conversation baseline DualTalkdyadic generation · Dyadic test set (Table 2)FD (FULL): 28.41 -> 10.43Improved
retrieval-based listener baselines vs ours on rotation and translation diversitylistener generation · Dyadic test set (Table 2)SID (LIS): 2.98 (Listen-R) vs 2.45No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy