Show HN: Supertonic 2 – 快速、设备端、多语言文本转语音
1 分•作者: login588•6 个月前
Supertonic 现已支持 5 种语言：
한국어 · Español · Français · Português · English
它旨在本地运行于各种设备上——从设备端/离线使用到通过 WebGPU 进行浏览器推理，以及移动设备和树莓派等小型设备。
以下是主要功能：
(1) 闪电般的速度——在 M4 Pro 上 RTF 0.006
(2) 轻量级——6600 万参数
(3) 设备端 TTS——完全隐私，零网络延迟
(4) 灵活部署——可在浏览器、PC、移动设备和边缘设备上运行
(5) 10 种预设声音——选择适合您用例的声音
(6) 开放权重模型——允许商业使用 (OpenRAIL-M)
我们希望 Supertonic 对许多应用有所帮助。
谢谢！
演示：<a href="https://huggingface.co/spaces/Supertone/supertonic-2" rel="nofollow">https://huggingface.co/spaces/Supertone/supertonic-2</a>
模型：<a href="https://huggingface.co/Supertone/supertonic-2" rel="nofollow">https://huggingface.co/Supertone/supertonic-2</a>
代码：<a href="https://github.com/supertone-inc/supertonic" rel="nofollow">https://github.com/supertone-inc/supertonic</a>
查看原文
Supertonic now supports 5 languages:
한국어 · Español · Français · Português · English<p>It’s built to run locally across setups—from on-device/offline use to browser inference with WebGPU, and small devices like mobile and Raspberry Pi.<p>Here are key features:<p>(1) Lightning fast — RTF 0.006 on M4 Pro<p>(2) Lightweight — 66M parameters<p>(3) On-device TTS — Complete privacy, zero network latency<p>(4) Flexible deployment — Runs on browsers, PCs, mobiles, and edge devices<p>(5) 10 preset voices — Pick the voice that fits your use cases<p>(6) Open-weight model — Commercial use allowed (OpenRAIL-M)<p>We hope Supertonic is useful for many applications.
Thank you!<p>Demo: <a href="https://huggingface.co/spaces/Supertone/supertonic-2" rel="nofollow">https://huggingface.co/spaces/Supertone/supertonic-2</a><p>Model: <a href="https://huggingface.co/Supertone/supertonic-2" rel="nofollow">https://huggingface.co/Supertone/supertonic-2</a><p>Code: <a href="https://github.com/supertone-inc/supertonic" rel="nofollow">https://github.com/supertone-inc/supertonic</a>