# WhaleRead WhaleRead is a local-first macOS workspace for translating and reading TXT, Markdown, and EPUB books across languages. It combines side-by-side bilingual reading, human-confirmed review, and optional Ask AI with a workflow that keeps core book work on the Mac. - Findings24: - Site: - Preços: Freemium - Categorias: Tradutor com IA ## O que é WhaleRead? WhaleRead is a local-first macOS workspace for translating and reading local books across languages. It supports TXT, Markdown, and EPUB files, with source, translated, and bilingual reading modes in the same workspace. Its core purpose is to keep translation, reading, review, and notes connected to the original title rather than sending an entire library into a consumer cloud. The public preview uses an on-device Hy-MT2 7B model for book translation on Apple silicon Macs running macOS 13 or later. The product also documents a self-hosted 30B translation path from its private evaluation setup, but that route is not included in the current download. Suggestions remain separate from the book until the reader reviews and explicitly applies them. An optional Ask AI workflow sends selected, visible context only to a compatible endpoint configured by the user. ## O que WhaleRead pode fazer? ### Local book translation Open TXT, Markdown, and EPUB files and translate them with the separately downloaded on-device Hy-MT2 7B model. The product site describes 18 common language directions. ### Source, translation, and bilingual views Switch among the original text, translated text, and a side-by-side bilingual layout while retaining chapter position, bookmarks, notes, typography, images, and in-book styling. ### Checkpointed reading workflow Read finished passages before the full task is complete and resume translation from verified checkpoints instead of waiting for an entire book conversion. ### Evidence-first human review Highlight a questionable passage and keep the source evidence, current wording, model reasoning, and proposed revision together. Reviewers can edit, keep, or discard a suggestion before applying it. ### Passage-based Ask AI Ask about an idiom, cultural reference, or character from within the reading page. The prepared context is shown before sending, the answer language is selectable, and useful responses can be kept with reading notes. ### Configurable model destination Core reading and editing stay on the Mac. Translation can use the public on-device 7B path, while the documented 30B path is self-hosted; Ask AI uses a compatible endpoint and credential configured by the user. ## Casos de uso ### Read a foreign-language book while translation is in progress Import an EPUB or another supported local file, read completed passages in bilingual mode, and continue from saved checkpoints as more of the book is translated. ### Improve literal or awkward passages Use the review record to compare the source and translation, inspect the model’s reasoning, rewrite a phrase when needed, and apply only confirmed changes to the reading edition. ### Study idioms and cultural references Select a passage from the page and ask a targeted question in the desired answer language, then save a useful explanation alongside reading notes without changing the book. ### Keep a private personal library workflow Work with local files and retain reading position, bookmarks, notes, and review history in the same macOS workspace. On-device translation keeps book text on the Mac. ## Perguntas frequentes ### Which files and platforms does WhaleRead support? The product supports local TXT, Markdown, and EPUB files. The public preview is for Apple silicon Macs running macOS 13 or later. ### Does translation run locally? The public download provides on-device Hy-MT2 7B translation, with the model downloaded separately. The site also describes a self-hosted 30B route from a private evaluation setup, but that route is not included in the current download. ### Can WhaleRead change a book without review? No. Model suggestions remain drafts until you inspect and explicitly apply them. The review workflow keeps the source evidence, current wording, reasoning, and proposed revision together. ### How does Ask AI handle book context? Ask AI shows the passage context prepared for the request before it is sent. Questions go only to the compatible endpoint and credential configured by the user, and answers do not edit the book automatically. ## Informações rápidas - **Product type**: Local-first book translation and bilingual reading workspace - **Platform**: Apple silicon Mac; macOS 13 or later - **Supported files**: TXT, Markdown, and EPUB - **Translation**: On-device Hy-MT2 7B in the public preview; 30B self-hosted route documented separately - **Language directions**: 18 common directions described on the product site - **Interface languages**: English and Chinese interface switching