Research map / 3D vision / Face and shape
EmoTalk: Speech-Driven Emotional Disentanglement for 3D Face Animation
Ziqiao Peng, Haoyu Wu, Zhenbo Song, Hao Xu, Xiangyu Zhu, Jun He, Hongyan Liu, Zhaoxin Fan
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 the emotion disentangling encoder (ablation) | EmoTalk · RAVDESS | EVE (mm): 3.076 -> 2.493 | Improved |
| add the emotion disentangling encoder (ablation) | EmoTalk · RAVDESS | LVE (mm): 3.126 -> 2.762 | Improved |
| add emotion-guided multi-head attention (ablation) | EmoTalk · RAVDESS | EVE (mm): 2.832 -> 2.493 | Improved |
| add velocity loss (ablation) | EmoTalk · RAVDESS | EVE (mm): 2.775 -> 2.493 | Improved |
| add classification loss on emotion extractor (ablation) | EmoTalk · RAVDESS | EVE (mm): 2.815 -> 2.493 | Improved |
| add HDTF data to training (ablation) | EmoTalk · RAVDESS | LVE (mm): 3.254 -> 2.762 | Improved |
| use wav2vec-based end-to-end encoder instead of Ji et al.'s MFCC staged encoder (ablation) | EmoTalk · RAVDESS | LVE (mm): 3.583 -> 2.762 | Improved |
| evaluate zero-shot on VOCA-Test without training on VOCASET | EmoTalk · VOCA-Test | LVE (mm): 4.418 -> 4.134 (vs FaceFormer) | Improved |
| EmoTalk vs FaceFormer on VOCASET lip average l2 error (zero-shot) | EmoTalk · VOCASET | lip average l2 error: 1.944 -> 1.914 | Improved |
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.
- MeshTalk — MeshTalk: 3D Face Animation From Speech Using Cross-Modality Disentanglement (ICCV 2021) · better on 3 of 3 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
- DEEPTalk: Dynamic Emotion Embedding for Probabilistic Speech-Driven 3D Face Animation (AAAI 2025 · similarity 0.91)
- Mimic: Speaking Style Disentanglement for Speech-Driven 3D Facial Animation (AAAI 2024 · similarity 0.91)
- Emotional Face-to-Speech (ICML 2025 · similarity 0.89)
- FaceChain-ImagineID: Freely Crafting High-Fidelity Diverse Talking Faces from Disentangled Audio (CVPR 2024 · similarity 0.89)
- Style2Talker: High-Resolution Talking Head Generation with Emotion Style and Art Style (AAAI 2024 · similarity 0.88)
- EmotiveTalk: Expressive Talking Head Generation through Audio Information Decoupling and Emotional Video Diffusion (CVPR 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.
