1 paper
Shun Lei, Yixuan Zhou, Liyang Chen +4
Previous works on expressive speech synthesis focus on modelling the mono-scale style embedding from the current sentence or context, but the multi-scale nature of speaking style i…