1 paper
Yizhong Geng, Yanliang Li, Jinghan Yang +3
Text-to-speech systems are increasingly used for drama dubbing, yet evaluation protocols remain sentence-level, leaving critical scene-level behaviors insufficiently measured. We p…