1 paper
Zhaolin Li, Jan Niehues
When building state-of-the-art speech translation models, the need for large computational resources is a significant obstacle due to the large training data size and complex model…