Read with Lexora

Start reading

Select a word, phrase, or passage on a web page where Lexora is enabled.

Read the result in context

Review the meaning or translation in the Lexora panel, then continue reading the original page. Available modules depend on the text, language, enabled features, and your settings.

Japanese readings and pronunciation

Use furigana when available to follow Japanese text. Pronunciation playback is available for supported languages; reading aids may vary with the selected text and settings.

Reading Tools · Word Length Highlighter

Open the Lexora toolbar → Lexora tab → Reading Tools. Enter a character count from 1 to 30 and press Highlight. Change the count and press Highlight to replace matches; use Clear when finished. Highlights remain during normal reading and supported dynamic-content changes. Matching is local and sends no page text to an AI provider. Restricted browser pages and some page content are unsupported.

Toolbar interface language

Choose Settings → Interface language. The toolbar, Reading Tools, YouTube controls, statuses, and errors follow that preference across the existing 10 interface languages. This setting is separate from translation source and target languages.

Save and return with Notebook

Explicitly save a selection from the reading panel, then find it in Notebook in Lexora Options. Entries stay in this browser. Search, delete, or export them as Markdown when you choose. Saving is unavailable in incognito tabs.

Translation methods and providers

In the Reading Panel, choose Machine Translation (Google Translate, online and the default), AI Translation (OpenAI, Gemini, or Claude), or On-device Translation (Chrome Translator). For AI Translation, use the separate provider control. Before switching, Lexora checks for that provider’s key in your chosen local or session storage. A missing key leaves the current route and result unchanged and shows guidance for that provider. Manage keys in Options → AI; Test connection sends a fixed test phrase to the selected provider. A successful route change translates the active selection again. Leaving AI Translation preserves the preferred AI provider. Translation uses only the selected provider and its key in the chosen storage mode; failures do not silently switch providers, models, or methods. Machine Translation and On-device Translation do not use AI credentials. On-device uses Chrome’s built-in Translator API with browser-managed language packs prepared in the panel. Availability depends on Chrome, the document, and the language pair; supported prepared pairs work offline.

These are the existing providers and engines. v0.21.0 adds no models, engines, local AI runtimes, or target languages. The selected provider is separate from the actual response model shown with an AI result.

Automatic reading and network requests

Wikipedia/Wikidata reference lookups, pronunciation, and Japanese reading may make network requests independently of translation mode. Automatic activation and pronunciation can trigger requests when enabled. Review those settings separately; choosing On-device Translation does not make every reading feature offline.

YouTube DualSubs

On supported YouTube videos, DualSubs displays original and translated captions. It is enabled by default and may load captions automatically from YouTube, including existing session cookies. Caption translation uses YouTube’s service independently of Reading Panel mode and AI-provider keys. Turn it off in settings or player controls. Adjust subtitle appearance and position live with the player controls.

If something is missing

Check page access, language pair, translation mode, and privacy settings in Options. For On-device Translation, follow the panel’s preparation progress or retry guidance; the API or pair may be unavailable. Online services need a connection. Contact Support with your installed version and steps tried, leaving out private text and credentials.