We’re open-sourcing *dots.tts* — the best voice cloning model. A 2B fully continuous, end-to-end AR TTS model with no discrete tokens, achieving top results on Seed-TTS-Eval and the best avg speaker…

Introducing dots.tts, an open-source end-to-end text-to-speech model from RedNote. dots.tts generates high-fidelity 48 kHz speech with a fully continuous architecture supporting style cloning from…

小红书发布了一个 2B 的 TTS模型 能实现文本转语音、声音克隆功能 支持24 种多语言 主要优势就是尺寸小，可以在端对端上部署