NVIDIA 发布 Magpie 多语言 TTS：开源权重，低延迟部署

AI 摘要

NVIDIA 发布 Magpie 多语言 TTS 模型，新增阿拉伯语、韩语和巴西葡萄牙语支持，共支持 12 种语言。该模型为 3.64 亿参数的开源权重模型，可通过 NVIDIA NIM 部署，在 B200 GPU 上实现 32 毫秒的首音频延迟和 320 倍实时吞吐。通过帧堆叠和局部 Transformer 架构优化，在保持语音质量的同时降低推理延迟，适用于多语言语音代理等场景。

正文节选

Build Low-Latency Multilingual Voice Agents: Open Weights & Full Deployment Control with NVIDIA Magpie TTS Every voice interaction has a latency budget. By the time a user hears your application respond, you've already spent precious milliseconds capturing audio, transcribing speech, running an LLM, retrieving context, and generating a response. Text-to-speech (TTS) is the final step — and the one users notice most. If speech generation is slow, the whole experience feels slow. The more of that

发布时间：—

抓取时间：2026-08-11 00:42

来源机构：Hugging Face