Investigating Pooling Strategies and Loss Functions for Weakly-Supervised Text-to-Audio Grounding via Contrastive Learning.

Xuenan Xu, Mengyue Wu, Kai Yu

Browse the full ICASSP paper archive.