AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

UniLS: End-to-End Audio-Driven Avatars for Unified Listening and Speaking

Xuangeng Chu, Ruicong Liu, Yifei Huang, Yun Liu, Yichen Peng, Bo Zheng

CVPR 2026 · Image generation › Autoregressive and latent generation · approach: Neural text-to-speech synthesis · 12 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 audio-free stage 1 pretraining of the motion prior (ablation)UniLS · Seamless Interaction listeningFDD: 25.64 -> 17.12Improved
add audio-free stage 1 pretraining (ablation)UniLS · Seamless Interaction listeningF-FID: 5.97 -> 4.304Improved
add audio-free stage 1 pretraining (ablation)UniLS · Seamless Interaction speakingLVE: 6.32 -> 5.83Improved
add diverse multi-scenario video data to stage 1 (ablation)UniLS · Seamless Interaction listeningF-FID: 4.62 -> 4.304Improved
add multi-scenario data to stage 1 (ablation)UniLS · Seamless Interaction speakingLVE: 6.26 -> 5.83Improved
use two separate cross-attention layers for speaker A and B audio instead of one on mixed audio (ablation)UniLS · Seamless Interaction speakingLVE: 11.48 -> 5.83Improved
use dual cross-attention instead of single cross-attention (ablation)UniLS · Seamless Interaction listeningFDD: 27.46 -> 17.12Improved
extend ARTalk to speak-listen by adding audio input (ARTalk*)ARTalk · Seamless Interaction listeningF-FID: 13.143 (DualTalk) vs 10.779 (ARTalk*)Improved
use UniLS end-to-end two-stage training instead of DualTalkconversational avatar · Seamless Interaction listeningP-FID: 0.079 -> 0.038Improved
jointly train speaking and listening directly from dual audio without a motion prior (ARTalk*)ARTalk · Seamless Interaction speakingPDD: 7.66 -> 8.55Hurt
use UniLS instead of ARTalk*conversational avatar · RTX 5090 inferenceFPS: 357.7 -> 560.6Improved

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