Token2vec: A Joint Self-Supervised Pre-Training Framework Using Unpaired Speech and Text.

Xianghu Yue, Junyi Ao, Xiaoxue Gao, Haizhou Li

Browse the full ICASSP paper archive.