Shared Labels

AbstractIt is well known that deep learning model has huge parameters and is computationally expensive, especially for embedded and mobile devices. Polyphone pronunciations selection is a basic function for Chinese …...