OnlyTranslate

Focused, capable, and intentionally quiet. It does one thing well: translation.

OnlyTranslate is an open-source browser extension for bilingual reading. It translates web content, video subtitles, local EPUB ebooks, and local or online PDFs while preserving the original structure and reading flow as much as possible. No OnlyTranslate account or subscription is required, and you choose the translation service.

If OnlyTranslate helps you, consider giving it a ⭐ Star to support the project and help more people discover an open-source bilingual reading tool.

Quick Start

Install OnlyTranslate from the Chrome Web Store, then open a regular webpage you want to translate.
Select the OnlyTranslate icon in the browser toolbar and choose an available translation service. No-setup services such as Microsoft Translator and Google Translate can be used immediately; AI services require their corresponding API keys.
For your first translation, try Bilingual + Smart, then select Translate current page. Select Restore original whenever you want to undo the translation.

To read an ebook or PDF, choose Reading and library (Beta) at the bottom of the popup and import a local DRM-free EPUB or PDF. You can also open an accessible online PDF with OnlyTranslate and save it to the library. Continue from the popup or use the full library to manage, back up, and restore books.

The More menu in the top-right of the popup contains Clear cache and Help. Help provides searchable instructions for translation modes, selection and hover translation, input fields, video subtitles, service setup, and common issues. The guide is bundled with the extension and is available offline.

Core Features

Smart / full-page translation: Smart mode focuses on articles, comments, and the main reading area. Full-page mode covers more visible text on documentation, forums, and tool pages.
Bilingual / translation-only display: Keep the original text for comparison and language learning, or show only the translation for a cleaner reading flow.
Multiple ways to translate: Start from the extension popup, the floating page toolbar, a keyboard shortcut, or the context menu.
Selection, hover, and input-field translation: Translate selected content or text under the pointer, preview suggestions beside the caret, switch target language in place, and press Tab to replace the original text.
Video subtitle translation: Captures source subtitles on supported sites, translates them in context-aware segments, and shows bilingual subtitles. Playback-aware scheduling and local caching reduce waiting and repeated requests.
EPUB ebook translation (Beta): Import local DRM-free EPUB books with automatic chapter translation, a table of contents, reading progress, bookmarks, and reading style controls.
PDF document translation (Beta): Read local or accessible online PDFs with page thumbnails, bilingual / translation-only / original-only modes, original-page comparison and zoom, and figure and formula display. An optional local layout model improves reading order on complex pages.
Flexible translation services: Includes presets for Microsoft Translator, Google Translate, Chrome's built-in translator, DeepL, OpenAI, DeepSeek, Gemini, Claude, and more, plus OpenAI Chat Completions-compatible gateways.
Advanced AI settings: Control thinking mode separately for each service and configure a default target language with an optional reverse-translation language.
Built-in user guide: Searchable offline instructions and interface screenshots are available in Simplified Chinese, English, Traditional Chinese, and Japanese.

Choosing a Translation Scope

Smart: Best for articles, blogs, posts, comments, and long-form reading. It prioritizes the main reading area and reduces noise from menus and navigation.
Full page: Best for documentation, admin pages, forums, tool sites, and information-dense pages. It translates more visible content while preserving the original page structure and interactions as much as possible.

If Smart mode misses content, switch to Full page and translate again. For dynamically loaded content, scroll it into view and trigger translation again.

Reading and Library (Beta)

Choose Reading and library at the bottom of the popup to import EPUB/PDF files, view recently read books, or continue reading. The full library supports book management, original-file export, and local library backup and restore.

EPUB Ebooks

Import a local DRM-free EPUB to automatically translate the current chapter as you read. The reader supports a table of contents, reading progress, position bookmarks, themes, font size, and line spacing.

PDF Documents

Open and save: Import a local PDF through Reading and library. When browsing an online PDF, choose Read this PDF with OnlyTranslate in the popup, or use the context menu on a PDF link. Online documents must be directly accessible; add them to the library to resume reading later.
Read and compare: Switch directly between bilingual, translation-only, and original-only modes in the toolbar. Jump between pages using thumbnails, collapse the original preview, or drag the divider to adjust the original and reading-flow panes. Themes, font size, and line spacing are configurable.
Papers and formulas: The reader attempts to reconstruct paragraphs that continue across columns or pages. Inline expressions support mathematical typesetting, while standalone equations and figures retain source images with double-click enlargement. Results on complex documents still depend on the original file structure.
Navigate and zoom: Use ← / → to change pages. Hold Ctrl / ⌘ and scroll the mouse wheel over the original page to zoom around the pointer.
Optional layout model: Basic layout analysis needs no additional download. For complex PDFs, use the layout model control in the reader's top toolbar to explicitly download the roughly 23 MB PP-DocLayout-M model. The verified model stays in private extension storage and can be removed at any time. Missing models, download failures, or runtime failures automatically fall back to basic analysis.
Files and limitations: Export the original PDF from the More menu. Translation requires extractable page text; image-only scans may not be translatable, and the optional layout model does not provide OCR text recognition.

Local Storage and Limitations

Local DRM-free EPUB and PDF files are supported, and accessible online PDFs can be saved to the library. MOBI and DRM/password-protected books are not supported yet.
Books, reading progress, and EPUB bookmarks are stored in the current browser profile. You can back up and restore them from the full library; cloud sync, notes, and AI analysis are not included.
Original EPUB/PDF files are not sent to the translation provider or model host. Extracted text is sent to the selected translation provider only when translation begins. Original-only PDF mode does not pre-send text for translation.
Removing a book also deletes its reading progress and bookmarks. Uninstalling the extension or clearing extension data removes the local library, so use Back up library first.
Imported books remain readable while the extension is disabled, but automatic translation pauses.

Supported Platforms

Web translation works on most regular webpages. Browser internal pages, extension stores, security-restricted pages, and some embedded content may prevent extensions from running.

Platform	Subtitle Translation	Web Translation
YouTube	Yes	Yes
Udemy	Yes	Yes
Coursera	Yes	Yes
Khan Academy	Yes	Yes
General webpages	—	Yes

Subtitle translation requires a readable source subtitle track provided by the video.

Privacy and Cost

OnlyTranslate itself is free and open source. It does not collect usage data for the project, and settings are stored locally in your browser.
When you use an online translation service, the text being translated is sent to the provider you selected and is subject to that provider's privacy policy.
Imported EPUB/PDF files, reading progress, bookmarks, and the optional PDF layout model remain in local browser storage. Only text that needs translation is sent to the selected provider.
Chrome's built-in translation runs locally in the browser, but availability depends on the Chrome version, language pair, and local model state.
Some online services require an API key and may charge according to their own terms. OnlyTranslate does not manage provider accounts, quotas, or billing.
Translations and video subtitles may be cached locally to reduce repeated requests. You can remove them with Clear cache in the popup's top-right More menu.

Installation

Chrome Web Store

OnlyTranslate - Chrome Web Store

Manual Installation

Download the latest .zip package from Releases and unzip it.
Open Chrome and go to chrome://extensions/.
Enable Developer mode in the top-right corner.
Select Load unpacked and choose the unzipped extension directory.

Development

# Install dependencies
corepack pnpm install

# Development mode for Chrome
corepack pnpm dev

# Type checking and the full test suite
corepack pnpm verify

# Build and package
corepack pnpm build
corepack pnpm zip

The project declares its pnpm version in package.json. Prefer corepack pnpm ... so scripts use the package manager version and local dependencies expected by the project.

For development mode, load .output/chrome-mv3-dev in Chrome developer mode. For production builds, load .output/chrome-mv3. Avoid mixing the two output directories.

Tech stack: WXT + Vue 3 + TypeScript, Manifest V3.

Feedback

If translation does not start, a page is only partially translated, or subtitles do not appear, check the built-in guide first. If the issue persists, report it through GitHub Issues.

Credits and Origin

OnlyTranslate is based on FluentRead. It keeps the core webpage translation experience while adding video subtitle translation, EPUB and PDF bilingual reading, Smart / Full page scope switching, and a more focused settings experience. Thanks to the original author and all contributors for their open-source work.

The design and implementation of OnlyTranslate's EPUB reader drew inspiration from taylorren/ai-reader. Thanks to the project and its contributors for sharing their open-source ideas and work.

PDF semantic layout analysis uses PaddleOCR's PP-DocLayout-M model and ONNX Runtime Web. The model is an optional, user-initiated download. See THIRD_PARTY_NOTICES.md for its source, checksum, and licensing information.

License

This project is open sourced under the GNU GPL v3.0 license.

FilesExpand file tree

README_EN.md

Latest commit

History