Inference-Time Scaling for Joint Audio-Video Generation

Arxiv CS.SD (Sound/Audio ML)

This paper introduces an inference-time scaling approach for joint audio-video generation, enabling synthesis of realistic, synchronized audio-video pairs from text without additional training. The method applies test-time compute scaling to enhance alignment and synchronization.

text-to-video

Arxiv CS.SD (Sound/Audio ML)·Jaemin JungKyeongha RhoInkyu ShinJoon Son Chung

8 days ago