1 paper
Hoyeon Lee, Sejung Son, Ye-Eun Kang +1
Current approaches to phrase break prediction address crucial prosodic aspects of text-to-speech systems but heavily rely on vast human annotations from audio or text, incurring si…