Is this Vietnamese text-to-speech really free?

Yes. The page is public, no signup is required, and there are no credit limits. The voice model runs entirely in your browser. For studio-grade voices and longer narration, see the paid Subformer Voice Synthesis service.

How does the Vietnamese TTS work in the browser?

It uses an open-source neural TTS engine compiled to WebAssembly and run on your own device. The voice model (~60 MB) is downloaded on first use and cached in your browser's Origin Private File System. Subsequent visits skip the download entirely.

Is my text uploaded to a server?

No. Synthesis runs locally in your browser. The only network calls are to download the voice model and the inference runtime - both are static assets. Your text and the generated audio never leave your device.

Does it work offline?

Once the voice model is cached, yes. The first visit needs network access to download the model; after that the page can synthesize speech offline.

What's the difference between this free Vietnamese TTS and the paid Voice Synthesis service?

The free in-browser TTS uses a medium-quality open-source voice - fine for narration drafts, accessibility, and quick previews, but it sounds noticeably synthetic. The paid Subformer Voice Synthesis service uses studio-grade neural models, supports voice cloning, and produces broadcast-quality audio. Visit /dashboard/voice-synthesis to try it.

Which browsers are supported?

Any modern browser with WebAssembly: Chrome, Firefox, Safari 15+, Edge, and Chromium-based mobile browsers. The page runs ORT in single-threaded mode so it works in every common browser without requiring cross-origin isolation.

Can I download the audio?

Yes. After synthesis you can play the audio inline or click Download to save it as a 22 kHz mono WAV file. Use any audio editor to convert to MP3 or other formats.

Subformer

免费 · 无需注册 · 在浏览器中运行

免费 Vietnamese 文本转语音

Tiếng Việt

在浏览器中输入或粘贴 Vietnamese 文本并听其朗读 - 无需注册。语音模型仅需下载一次（~60 MB）并缓存于本地，后续运行即时且支持离线。

工作原理

1. 输入文本
粘贴或输入任意 Vietnamese 文本 - 标点和换行将被保留。
2. 生成
首次点击会将~60 MB 的语音模型下载到您的浏览器。之后，合成即时完成。
3. 下载或播放
在页面内播放音频或将其保存为 WAV 文件。音频永远不会离开您的设备。

您的文本保留在您的设备上

合成完全在您的浏览器中使用WebAssembly运行。唯一的网络请求是获取语音模型和推理运行时——两者都是公共的静态资产。我们看不到您的文本，也看不到生成的音频。

Vietnamese语音如何在您的浏览器中运行

这是一个在您自己的机器上执行的真实神经 TTS 模型，而不是托管 API 的瘦客户端。有四个部分使其在标签页中成为可能。

Open-source neural voice: Vietnamese模型在单个说话者上训练，这就是为什么您合成的所有内容中的口音和语调都保持一致的原因。
WebAssembly: 模型运行时和将Tiếng Việt文本转换为音素的步骤都编译为WebAssembly，因此最初为桌面编写的代码可以在浏览器沙盒中以接近原生速度运行。
On-device inference: 模型的神经网络在本地执行，固定为单线程模式，因此页面无需跨域隔离，并可在所有现代浏览器中运行。
Origin Private File System: 首次运行后，约 60 MB 的语音模型会缓存在私有的、按来源存储的空间中。后续访问将跳过下载，无需网络即可合成。

需要广播级音色吗？