It just dropped. Apache licence. Two models. FireRedTTS3-Base, zero-shot voice cloning across 24 languages and 21 Chinese dialects. FireRedTTS3-Instruct, natural-language voice design and speech editing (semantic + acoustic) in one unified model.
https://github.com/FireRedTeam/FireRedTTS3
It just dropped. Apache licence. Two models. FireRedTTS3-Base, zero-shot voice cloning across 24 languages and 21 Chinese dialects. FireRedTTS3-Instruct, natural-language voice design and speech editing (semantic + acoustic) in one unified model.
https://github.com/FireRedTeam/FireRedTTS3