Research map / Video / Video generation
Contrastive Decoupled Representation Learning and Regularization for Speech-Preserving Facial Expression Manipulation
Tianshui Chen, Jianman Lin, Zhijing Yang, Chunmei Qing, Yukai Shi, Liang Lin
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 |
|---|---|---|---|
| integrate CDRL supervision into NED | NED · MEAD intra-ID | average FAD: 2.108 -> 1.026 | Improved |
| integrate CDRL supervision into NED | NED · MEAD intra-ID | average CSIM: 0.831 -> 0.914 | Improved |
| integrate CDRL supervision into NED | NED · MEAD cross-ID | average LSE-D: 9.906 -> 9.351 | Improved |
| integrate CDRL into NED without retraining CDRL | NED · RAVDESS cross-ID | average FAD: 5.412 -> 4.816 | Improved |
| add CCRL content supervision (ablation) | Ours (NED) · MEAD cross-ID | LSE-D: 9.883 -> 9.351 | Improved |
| add audio as content prior in CCRL (ablation) | CCRL · MEAD cross-ID | LSE-D: 9.812 -> 9.351 | Improved |
| add emotion-aware negatives in CCRL contrastive loss (ablation) | CCRL · MEAD cross-ID | LSE-D: 9.382 -> 9.351 | Improved |
| add CERL emotion supervision (ablation) | Ours (NED) · MEAD intra-ID | CSIM: 0.837 -> 0.914 | Improved |
| add CERL emotion supervision (ablation) | Ours (NED) · MEAD cross-ID | CSIM: 0.774 -> 0.792 | Improved |
| use dot product instead of cross-attention in CERL (ablation) | CERL · MEAD intra-ID | CSIM: 0.899 -> 0.914 | Improved |
| filter emotionally expressive images with CLIP for CERL training (emotion-augmented contrastive learning) (ablation) | CERL · MEAD intra-ID | CSIM: 0.888 -> 0.914 | Improved |
| combine ASCCL with CDRL (ablation) | CDRL on NED · MEAD cross-ID | FAD: 4.344 -> 4.252 | Improved |
Closest papers by meaning
- Learning Adaptive Spatial Coherent Correlations for Speech-Preserving Facial Expression Manipulation (CVPR 2024 · similarity 0.87)
- EmotiveTalk: Expressive Talking Head Generation through Audio Information Decoupling and Emotional Video Diffusion (CVPR 2025 · similarity 0.87)
- Emotional Face-to-Speech (ICML 2025 · similarity 0.85)
- DEEPTalk: Dynamic Emotion Embedding for Probabilistic Speech-Driven 3D Face Animation (AAAI 2025 · similarity 0.85)
- Cafe-Talk: Generating 3D Talking Face Animation with Multimodal Coarse- and Fine-grained Control (ICLR 2025 · similarity 0.84)
- Emotion Rendering for Conversational Speech Synthesis with Heterogeneous Graph-Based Context Modeling (AAAI 2024 · similarity 0.84)
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.
