AV-FlowEdit: Training-Free Prompt-Driven Audio-Visual Editing with Event-Aware Spectral Trajectory Preservation
We present AV-FlowEdit, a training-free and inversion-free framework for prompt-driven audio-video editing.
ProofPaper ↗
Key points
- Joint audio-video diffusion models generate synchronized content, but editing existing clips requires target changes without disrupting non-target structure or cross-modal correspondence.
- Our key insight is that source--target flow differences provide both cross-modal editing directions and preservation cues.
- To suppress drift accumulated during sampling, Cross-Modal Event-Aware Spectral Trajectory Preservation combines flow saliency, aligned audio activity, multiscale low-frequency displacement, and diffusion-stage scheduling.
- Experiments show strong prompt alignment and temporal consistency, together with competitive audio quality and cross-modal coherence.
Sources (1)
- [1]AV-FlowEdit: Training-Free Prompt-Driven Audio-Visual Editing with Event-Aware Spectral Trajectory PreservationarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 06:31 AM
We present AV-FlowEdit, a training-free and inversion-free framework for prompt-driven audio-video editing.
Joint audio-video diffusion models generate synchronized content, but editing existing clips requires target changes without disrupting non-target structure or cross-modal correspondence.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 9, 2026Qwen/Qwen-Image-2.1-Turbo
- Oct 9, 2026Learning Representations to Guide Pixel Diffusion Transformers
- Oct 8, 2026Shift-Aware Identity-Guided Latent Refinement for Referring Audio--Visual Segmentation
- Oct 8, 2026StereoEngine: Scaling Stereo Bimanual Robot Data without Stereo Cameras
- Oct 8, 2026LEGO: A Lifting-Free Approach for Exocentric-to-Egocentric Video Generation
- Oct 8, 2026DiffusionART: Efficient, Stable and Scalable Reinforcement of Diffusion Models