Research map / Video / Video generation
StyleTalk: One-Shot Talking Head Generation with Controllable Speaking Styles
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 style-aware adaptive feed-forward layers instead of vanilla feed-forward layers (ablation) | StyleTalk · MEAD | F-LMD: 2.414 -> 2.122 | Helped |
| add style-aware adaptive feed-forward layers instead of vanilla feed-forward layers (ablation) | StyleTalk · MEAD | M-LMD: 4.178 -> 3.249 | Helped |
| add style-aware adaptive feed-forward layers instead of vanilla feed-forward layers (ablation) | StyleTalk · MEAD | Sync_conf: 3.059 -> 3.474 | Helped |
| set adaptive feed-forward kernel count to 4 rather than 8 (ablation) | StyleTalk · MEAD | F-LMD: 2.122 -> 2.327 | Hurt |
| set adaptive feed-forward kernel count to 4 rather than 8 (ablation) | StyleTalk · MEAD | M-LMD: 3.249 -> 3.524 | Hurt |
| set adaptive feed-forward kernel count to 16 rather than 8 (ablation) | StyleTalk · MEAD | F-LMD: 2.122 -> 2.133 | No clear effect |
| set adaptive feed-forward kernel count to 16 rather than 8 (ablation) | StyleTalk · MEAD | M-LMD: 3.249 -> 3.396 | Hurt |
| add the style discriminator (ablation) | StyleTalk · MEAD | F-LMD: 2.483 -> 2.122 | Helped |
| add the style discriminator (ablation) | StyleTalk · MEAD | M-LMD: 3.628 -> 3.249 | Helped |
| add the triplet loss (ablation) | StyleTalk · MEAD | F-LMD: 2.401 -> 2.122 | Helped |
| add the triplet loss (ablation) | StyleTalk · MEAD | M-LMD: 3.771 -> 3.249 | Helped |
| add the lip-sync discriminator (ablation) | StyleTalk · MEAD | M-LMD: 4.351 -> 3.249 | Helped |
Closest papers by meaning
- Say Anything with Any Style (AAAI 2024 · similarity 0.91)
- Style2Talker: High-Resolution Talking Head Generation with Emotion Style and Art Style (AAAI 2024 · similarity 0.91)
- 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)
- Synergizing Motion and Appearance: Multi-Scale Compensatory Codebooks for Talking Head Video Generation (CVPR 2025 · similarity 0.88)
- MimicTalk: Mimicking a personalized and expressive 3D talking face in minutes (NeurIPS 2024 · 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.