Investigating the Utility of Surprisal from Large Language Models for Speech Synthesis Prosody

Open in new window