1 paper
Hongxuan Lu, Shenjian Wang, Biao Li
During the entire training process of the ASR model, the intensity of data augmentation and the approach of calculating training loss are applied in a regulated manner based on pre…