SGLang-Omni 刚上线 MOSS-TTS v1.5，开源、零样本克隆声音，支持31种语言，速度也不错，玩玩看。
SGLang-Omni 现已支持 MOSS-TTS-Local Transformer v1.5 模型。该模型基于 Qwen3-4B 骨干，可生成 48kHz 立体声语音。支持零样本语音克隆和原生流式，覆盖 31 种语言，训练数据约 400 万小时。非流式场景下达到 5.976 req/s，RTF 0.644，WER 1.75%（SeedTTS English，2×GPU）。采用三阶段管线：参考编码、AR 引擎、流式声码器。
🎉 SGLang-Omni now serves MOSS-TTS-Local Transforme…
🎉 SGLang-Omni now serves MOSS-TTS-Local Transformer v1.5 from @Open_MOSS on day 0! This is an open 48 kHz stereo TTS model built on a Qwen3-4B backbone. ✅ Zero-shot voice cloning + native streaming at 48 kHz stereo ✅ 31 languages, trained on ~4M hours of speech ✅ Duration control + explicit pause markup + long-form up to 10 min ✅ 5.976 req/s non-streaming at RTF 0.644, 1.75% WER (SeedTTS English, 2× GPU) ✅ Three-stage pipeline: reference encoding → AR engine → streaming vocoder, with frame-level CUDA Graphs
Cookbook: https://t.co/FuxkTbFkZH Run it now with SGLang-Omni!