Reproduce “E3-System: Expressive EmoData, Explicitly Controllable, Easy-to-Use TTS” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yiqiao Huang, Xiaolong Lin, Zhenpeng Zhan, Zhizheng Wu
E3-System addresses expressive text-to-speech by retrieving speech that acoustically grounds a style and using it separately from the target-speaker prompt. Style-RAG matches descriptions to speech clips, and Dual-TTS combines the style and speaker references. In LibriTTS tests, the no-retrieval system scores 70.15 SIM-Spk and 89.22 SIM-Style, versus 59.02 and 72.50 with GRL, supporting the training strategy in this setting, not universal separation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.