Learning Self-Supervised Audio-Visual Representations for Sound Recommendations

Sudha Krishnamurthy

We propose a novel self-supervised approach for learning audio and visual representations from unlabeled videos, based on their correspondence. The approach uses an attention mecha…