WebNarrator 作者： David Reid
Select any text on any webpage, hit 'play', hear it read aloud with live word-by-word highlighting. Uses your OS's installed voices. No cloud, no tracking, no data collected, works offline.
17 个用户17 个用户
扩展元数据
屏幕截图
关于此扩展
WebNarrator is a privacy-first text-to-speech reader that uses your operating system's built-in voices, or its own bundled on-device neural voices. No analytics, no telemetry, no tracking. The only network event is an optional one-time download of the Kokoro model from HuggingFace if you turn on neural voices.
As of version 1.1.0, WebNarrator ships with the ability to download the Kokoro-82M neural voice model (~310 MB, one-time download), which adds six high-quality on-device voices (3 female, 3 male, US and UK English). Downloading Kokoro is optional, and once downloaded it runs locally on your machine.
How to use it
Select any text on any webpage.
Press Alt + Shift + S (or ⌥ + ⇧ + S on macOS) or right-click and choose "Read selected text aloud".
The current word highlights in yellow as it's spoken.
The toolbar popup gives you play / pause / stop controls, a 0.5× to 3× speed slider, and a voice picker for any TTS voice your OS has installed, plus WebNarrator's optional neural voices.
What makes it different
• Live word-level highlighting: see exactly what's being spoken, on the page itself, not in a separate sidebar.
• Works everywhere: articles, docs, blog posts, social media, code-heavy pages, math notation (KaTeX / MathJax), and iframe-embedded content (LeetCode, embedded readers, etc.).
• Customizable keyboard shortcut: configure at about:addons → Manage Extension Shortcuts.
• Light and dark themes: follows your OS preference automatically, or pick manually.
• No internet required: speech is generated by your OS, on your device. Neural voices work fully offline once the Kokoro model has been downloaded the first time. Disconnect from Wi-Fi and WebNarrator works exactly the same.
Privacy: no data collected
WebNarrator collects, stores, and transmits zero user data. No analytics, no telemetry, no tracking pixels. The only things saved locally are your voice choice, your speech rate, your light/dark theme preference, and the cached neural model (if you choose to use it), all stored in your browser, never sent anywhere.
Full privacy policy: https://webnarrator.com/privacy/
A note on voice quality
WebNarrator uses whatever voices your operating system has installed. Most operating systems ship with one or two basic voices and require you to download the high-quality "premium" or "neural" voices separately. If WebNarrator sounds robotic out of the box, you almost certainly have only the basic voices installed. The fastest fix is to turn on WebNarrator's built-in neural voices (the Kokoro model) in the popup. If you'd rather use OS voices, here is a step-by-step guide for macOS, Windows, Linux, and ChromeOS: https://webnarrator.com/posts/download-better-voices/
As of version 1.1.0, WebNarrator ships with the ability to download the Kokoro-82M neural voice model (~310 MB, one-time download), which adds six high-quality on-device voices (3 female, 3 male, US and UK English). Downloading Kokoro is optional, and once downloaded it runs locally on your machine.
How to use it
Select any text on any webpage.
Press Alt + Shift + S (or ⌥ + ⇧ + S on macOS) or right-click and choose "Read selected text aloud".
The current word highlights in yellow as it's spoken.
The toolbar popup gives you play / pause / stop controls, a 0.5× to 3× speed slider, and a voice picker for any TTS voice your OS has installed, plus WebNarrator's optional neural voices.
What makes it different
• Live word-level highlighting: see exactly what's being spoken, on the page itself, not in a separate sidebar.
• Works everywhere: articles, docs, blog posts, social media, code-heavy pages, math notation (KaTeX / MathJax), and iframe-embedded content (LeetCode, embedded readers, etc.).
• Customizable keyboard shortcut: configure at about:addons → Manage Extension Shortcuts.
• Light and dark themes: follows your OS preference automatically, or pick manually.
• No internet required: speech is generated by your OS, on your device. Neural voices work fully offline once the Kokoro model has been downloaded the first time. Disconnect from Wi-Fi and WebNarrator works exactly the same.
Privacy: no data collected
WebNarrator collects, stores, and transmits zero user data. No analytics, no telemetry, no tracking pixels. The only things saved locally are your voice choice, your speech rate, your light/dark theme preference, and the cached neural model (if you choose to use it), all stored in your browser, never sent anywhere.
Full privacy policy: https://webnarrator.com/privacy/
A note on voice quality
WebNarrator uses whatever voices your operating system has installed. Most operating systems ship with one or two basic voices and require you to download the high-quality "premium" or "neural" voices separately. If WebNarrator sounds robotic out of the box, you almost certainly have only the basic voices installed. The fastest fix is to turn on WebNarrator's built-in neural voices (the Kokoro model) in the popup. If you'd rather use OS voices, here is a step-by-step guide for macOS, Windows, Linux, and ChromeOS: https://webnarrator.com/posts/download-better-voices/
评分 0（1 位用户）
权限与数据
更多信息