Machine Heart
Aug 20, 2026 · Artificial Intelligence
FireRedTTS3: Zero‑Shot Multilingual Voice Cloning, Design, and Editing in One Model
FireRedTTS3 unifies 24‑language, 21‑dialect zero‑shot voice cloning, natural‑language voice design, and precise local editing in a single model by injecting semantic information into a continuous speech representation (RedAE) and leveraging a lightweight LLM‑DiT generation framework, achieving state‑of‑the‑art results on four public benchmarks.
LLM-DiTMultilingual TTSRedAE
0 likes · 13 min read
