Research map / Video / Video generation
CoordSpeaker: Exploiting Gesture Captioning for Coordinated Caption-Empowered Co-Speech Gesture Generation
Fengyi Fang, Sicheng Yang, Wenming Yang
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 multi-granular captioning (ablation) | CoordSpeaker · BEAT + HumanML3D joint control | MM-Dist / R-Precision: MM-Dist 7.031 -> 6.814; R-Precision 0.082 -> 0.100 | Improved |
| add hierarchically controlled denoiser (ablation) | CoordSpeaker · BEAT + HumanML3D joint control | Beat Consistency (x10^-1): 1.910 -> 1.327 (w/o hcd. to full; BC closer to co-speech-only bias removed) | Improved |
| add hierarchically controlled denoiser (ablation) | CoordSpeaker · BEAT + HumanML3D joint control | Jerk: 1.201 -> 1.190 | Improved |
| add HumanML3D with unified cross-dataset motion representation (ablation) | CoordSpeaker · BEAT + HumanML3D joint control | FGD: 19.600 -> 10.861 | Improved |
| add gesture captions (ablation) | CoordSpeaker · user study, matching | win rate (%): 20.0 -> 33.0 | Improved |
| add hierarchical captions with regular local captions (ablation) | CoordSpeaker · BEAT + HumanML3D | overall balance | Improved |
| use hierarchical condition injection with latent diffusion instead of FreeTalker | CoordSpeaker · inference time per sentence | AITS (s): 6.632 -> 0.842 | Improved |
| use CoordSpeaker versus FreeTalker under joint control | CoordSpeaker · BEAT joint control | FGD: BC 1.147 -> 1.327; FGD 11.332 -> 10.861; MM-Dist 6.737 -> 6.814 | Mixed |
| use CoordSpeaker versus SynTalker on text-to-motion | CoordSpeaker · HumanML3D test | FID: 4.385 -> 0.405 | Improved |
| compare with MoMask on text-only generation | CoordSpeaker · HumanML3D test | FID: 0.222 (MoMask) vs 0.405 | Hurt |
Closest papers by meaning
- Co$^{\mathbf{3}}$Gesture: Towards Coherent Concurrent Co-speech 3D Gesture Generation with Interactive Diffusion (ICLR 2025 · similarity 0.92)
- ConvoFusion: Multi-Modal Conversational Diffusion for Co-Speech Gesture Synthesis (CVPR 2024 · similarity 0.92)
- DiffSHEG: A Diffusion-Based Approach for Real-Time Speech-driven Holistic 3D Expression and Gesture Generation (CVPR 2024 · similarity 0.90)
- Co-Speech Gesture Video Generation via Motion-Decoupled Diffusion Model (CVPR 2024 · similarity 0.90)
- Co-Speech Gesture Video Generation with Implicit Motion-Audio Entanglement (CVPR 2025 · similarity 0.90)
- Retrieving Semantics from the Deep: an RAG Solution for Gesture Synthesis (CVPR 2025 · similarity 0.89)
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.
