Everything you need for a long read
Read a whole webpage, a selected passage, or a text-based PDF. PageNarrator clears away page clutter, keeps the current passage visible, remembers your position, and lets you listen from 0.5× to 3× speed.
Every feature is included. There is no PageNarrator account, subscription, or usage quota.
Built-in voices do not upload your reading
Missing voice files download from Hugging Face when you prepare a voice or start reading. Speech then runs inside the extension. Successfully cached files can be reused offline for the same voice and processing mode, unless the cache is cleared or evicted. PageNarrator does not send the text, URL, PDF, or browsing history to its servers for narration.
Try the 1.0.0 preview
Version 1.0.0 is being prepared for release. The repository is currently private and the Chrome Web Store listing is not available yet. Public availability is pending.
Repository access is required. If you do not have access, contact support about the preview.
Build and install the preview
These steps are for testers who already have repository access. Install Node.js 22.12 or newer, then:
- Clone the repository using the HTTPS URL from its GitHub Code menu.
- Open a terminal in the cloned folder:
cd PageNarrator. - Install the dependencies:
npm ci. - Build the extension:
npm run build. - Open
chrome://extensions, enable Developer mode, choose Load unpacked, and select thedist/folder.
For local PDFs, enable Allow access to file URLs in the extension's details. Preparing a voice requires an initial download.
What your computer needs
Use a desktop Chromium 120+ browser. Kokoro supports WebGPU and CPU-only WebAssembly; CPU setup and speech take longer. Experimental OmniVoice requires WebGPU with fp16 support and a separate first-use download of about 1.35 GB. Performance depends on your device and document.