1 paper · 1 filter
Zeqian Ju, Yuancheng Wang, Kai Shen +16
While recent large-scale text-to-speech (TTS) models have achieved significant progress, they still fall short in speech quality, similarity, and prosody. Considering speech intric…