acappemin / video-to-audio-and-piano

Enhance Generation Quality of Flow Matching V2A Model via Multi-Step CoT-Like Guidance and Combined Preference Optimization

• Public
• 24 runs
• GitHub
• Weights
• Paper
• License