Masked Autoencoders with Multi-Window Local-Global Attention Are Better Audio Learners.

Sarthak Yadav, Sergios Theodoridis, Lars Kai Hansen, Zheng-Hua Tan

Browse the full ICLR paper archive.