结合音频模型测试音效与画面匹配度

I noticed that music is less present in Seedance 2.5 outputs compared to Seedance 2.0. So for this video, once I finished the full edit, I uploaded it into a .skills and asked it to analyze the sequences and create a prompt for the sound design. Then I used that prompt in Seed Audio 1.0. The result is actually pretty good.