Research map / Video / Video generation
JoVA: Unified Multimodal Learning for Joint Video-Audio Generation and Editing
Xiaohu Huang, Haoyang He, Hao Zhou, Qiangpeng Yang, Min Zheng, Kai Han
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 mouth-area loss (lambda 0.0 to 2.0) (ablation) | JoVA · JoVABench-Gen | LSE-C: 1.39 -> 6.53 | Improved |
| raise mouth-area loss weight from 2.0 to 5.0 (ablation) | JoVA · JoVABench-Gen | LSE-C: 6.53 -> 6.70 | Improved |
| raise mouth-area loss weight from 5.0 to 8.0 (ablation) | JoVA · JoVABench-Gen | LSE-C: 6.70 -> 6.65 | Hurt |
| vary mouth-area loss weight lambda (ablation) | JoVA · JoVABench-Gen | PQ: 6.44, 6.47, 6.49, 6.40 | No clear effect |
| use joint self-attention instead of sequential Ovi-style cross-attention (ablation) | JoVA · JoVABench-Gen | LSE-C: 6.51 -> 6.70 | Improved |
| use joint self-attention instead of sequential Ovi-style cross-attention (ablation) | JoVA · JoVABench-Gen | WER: 0.24 -> 0.19 | Improved |
| use joint self-attention instead of parallel cross-attention with linear adaptation (ablation) | JoVA · JoVABench-Gen | LSE-C: 1.59 -> 6.70 | Improved |
| use joint self-attention instead of parallel cross-attention without linear (ablation) | JoVA · JoVABench-Gen | PQ: 6.14 -> 6.49 | Improved |
| use JoVA unified model instead of Ovi on Verse-Bench | JoVA · Verse-Bench | DeSync: 0.49 -> 0.16 | Improved |
| use JoVA instead of best cascaded editing baseline (Ditto + Wav2Lip) | JoVA · JoVABench-Edit | LSE-D: 2.83 -> 1.66 | Improved |
| use JoVA instead of Ditto + Wav2Lip in human study | JoVA · Video-audio editing MOS | Instruction-Following: 3.30 -> 3.85 | Improved |
| use JoVA instead of Gen-3 on pure video generation | JoVA · VBench | Total Score: 82.32 -> 82.61 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- UniAVGen — UniAVGen: Unified Audio and Video Generation with Asymmetric Cross-Modal Interactions (CVPR 2026) · better on 3 of 4 shared results
- JavisDiT — JavisDiT: Joint Audio-Video Diffusion Transformer with Hierarchical Spatio-Temporal Prior Synchronization (ICLR 2026) · better on 1 of 1 shared result
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
- MMAudio: Taming Multimodal Joint Training for High-Quality Video-to-Audio Synthesis (CVPR 2025 · similarity 0.89)
- Native Audio-Visual Alignment for Generation (NeurIPS 2026 · similarity 0.88)
- Diverse and Aligned Audio-to-Video Generation via Text-to-Video Model Adaptation (AAAI 2024 · similarity 0.87)
- JavisGPT: A Unified Multi-modal LLM for Sounding-Video Comprehension and Generation (NeurIPS 2025 · similarity 0.87)
- Emu: Generative Pretraining in Multimodality (ICLR 2024 · similarity 0.87)
- Model-Guided Dual-Role Alignment for High-Fidelity Open-Domain Video-to-Audio Generation (NeurIPS 2025 · similarity 0.87)
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.
