Research map / Vision-language and multimodal / CLIP and zero-shot
InstructDubber: Instruction-based Alignment for Zero-shot Movie Dubbing
Zhedong Zhang, Liang Li, Gaoxiang Cong, Chunshan Liu, Yuhan Gao, Xiaowan Wang, Tao Gu, Yuankai Qi
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 IDD (instructed duration distilling) to visual-feature baseline (ablation) | visual-feature dubbing baseline · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | DD: 0.5030 -> 0.4941 | Improved |
| add IDD to visual-feature baseline (ablation) | visual-feature dubbing baseline · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | WER (%): 10.54 -> 9.97 | Improved |
| add IDD to visual-feature baseline (ablation) | visual-feature dubbing baseline · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | EMO-SIM (%): 69.51 -> 67.85 | Hurt |
| add IEC (instructed emotion calibrating) to visual-feature baseline (ablation) | visual-feature dubbing baseline · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | EMO-SIM (%): 69.51 -> 70.77 | Improved |
| add IEC to visual-feature baseline (ablation) | visual-feature dubbing baseline · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | WER (%): 10.54 -> 11.76 | Hurt |
| use both IDD and IEC (ablation) | InstructDubber · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | DD: 0.5030 -> 0.4933 | Improved |
| use IDD slot distillation instead of raw instruction cross-attention (ablation) | InstructDubber · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | DD: 0.5072 -> 0.4933 | Improved |
| use IDD instead of LoRA LLM duration prediction (ablation) | InstructDubber · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | DD: 0.5331 -> 0.4933 | Improved |
| use 10 duration prototypes instead of 1 (ablation) | IDD · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | DD: 0.5005 -> 0.4933 | Improved |
| use 20 duration prototypes instead of 10 (ablation) | IDD · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | DD: 0.4933 -> 0.4991 | No clear effect |
| add ground-truth emotion entity calibration of the analyzer (ablation) | IEC · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | EMO-SIM (%): 70.33 -> 70.94 | Improved |
| use calibrated entity analysis instead of raw instruction cross-attention (ablation) | IEC · Ablation avg of V2C-Animation, Chem, V2C2Chem, Chem2V2C | EMO-SIM (%): 69.55 -> 70.94 | Improved |
Closest papers by meaning
- Prosody-Enhanced Acoustic Pre-training and Acoustic-Disentangled Prosody Adapting for Movie Dubbing (CVPR 2025 · similarity 0.91)
- EmoDubber: Towards High Quality and Emotion Controllable Movie Dubbing (CVPR 2025 · similarity 0.91)
- ProsodyTalker: 3D Visual Speech Animation via Prosody Decomposition (AAAI 2025 · similarity 0.85)
- DiVAS: Video and Audio Synchronization with Dynamic Frame Rates (CVPR 2024 · similarity 0.85)
- An eye for an ear: zero-shot audio description leveraging an image captioner with audio-visual token distribution matching (NeurIPS 2024 · similarity 0.84)
- InstructSpeech: Following Speech Editing Instructions via Large Language Models (ICML 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.
