Launching today
Confucius4
a multilingual and cross-lingual zero-shot TTS engine
2 followers
a multilingual and cross-lingual zero-shot TTS engine
2 followers
Confucius4-TTS is an advanced, LLM-based text-to-speech engine built for seamless multilingual and cross-lingual synthesis. Supporting 14 languages, it features zero-shot voice cloning with no reference transcript required. Unlike traditional TTS, it delivers flawless cross-lingual voice transfer—allowing any speaker identity to switch languages with natural emotion, zero accent, and robust generalization.
HeyForm
Thrilled to hunt Confucius4 today.
Developed by NetEase Youdao, this advanced LLM-based engine tackles one of the biggest headaches in global content creation: making cross-lingual voice cloning actually sound natural. It supports 14 languages (including English, Chinese, Japanese, and Spanish) and allows you to perfectly replicate a speaker's voice identity and emotion across different languages without a reference transcript or heavy training—all while dropping the native accent entirely.
Check it out and let the team know your thoughts or how you plan to use it for your multilingual projects!