Research map / Video / Video generation
READ: Real-time and Efficient Asynchronous Diffusion for Audio-driven Talking Head Generation
Haotian Wang, Yuzhe Weng, Jun Du, Haoran Xu, Xiaoyan Wu, Shan He, Bing Yin, Cong Liu, Jianqing Gao, Qingfeng Liu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add self-supervised pre-training of SpeechAE (ablation) | READ · HDTF | Sync-C: 7.965 -> 8.658 | Improved |
| add self-supervised pre-training of SpeechAE (ablation) | READ · HDTF | Sync-D: 7.361 -> 6.890 | Improved |
| add self-supervised pre-training of SpeechAE (ablation) | READ · HDTF | FID: 15.305 -> 15.073 | Improved |
| add SpeechAE for temporal compression of speech features (ablation) | READ · HDTF | Sync-C: 2.086 -> 8.658 | Improved |
| add SpeechAE for temporal compression of speech features (ablation) | READ · HDTF | Sync-D: 12.415 -> 6.890 | Improved |
| add SpeechAE for temporal compression of speech features (ablation) | READ · HDTF | FID: 15.617 -> 15.073 | Improved |
| add asynchronous noise scheduler (ANS) instead of synchronous noise and clip concatenation (ablation) | READ · two consecutive 121-frame clips | temporal consistency (qualitative) | Improved |
| extend generated video length from 121 to 457 and 1017 frames (ablation) | READ with ANS · HDTF | Sync-C: 8.658 -> 8.767 -> 8.677 | No clear effect |
| extend generated video length from 121 to 1017 frames (ablation) | READ with ANS · HDTF | FID: 15.073 -> 15.241 -> 15.195 | No clear effect |
| use Split-CFG instead of Joint-CFG (ablation) | READ · HDTF | lip-sync and runtime: runtime increased almost by 50% | Mixed |
| reduce inference steps below 5 (ablation) | READ · HDTF | FID | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Sonic — Sonic: Shifting Focus to Global Audio Perception in Portrait Animation (CVPR 2025) · better on 9 of 13 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
- MoDiTalker: Motion-Disentangled Diffusion Model for High-Fidelity Talking Head Generation (AAAI 2025 · similarity 0.89)
- OmniTalker: One-shot Real-time Text-Driven Talking Audio-Video Generation With Multimodal Style Mimicking (NeurIPS 2025 · similarity 0.89)
- EmotiveTalk: Expressive Talking Head Generation through Audio Information Decoupling and Emotional Video Diffusion (CVPR 2025 · similarity 0.89)
- Diffusion-based Realistic Listening Head Generation via Hybrid Motion Modeling (CVPR 2025 · similarity 0.88)
- FaceTalk: Audio-Driven Motion Diffusion for Neural Parametric Head Models (CVPR 2024 · similarity 0.88)
- DAWN: Dynamic Frame Avatar with Non-autoregressive Diffusion Framework for Talking head Video Generation (ICLR 2025 · similarity 0.88)
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.
