Immersive Translate

Immersive Translate's differentiator is structural, not promotional: the source text stays on the page and the translation sits next to it, paragraph by paragraph. That design choice is why researchers reading papers and developers scanning foreign release notes prefer it over a single-language overlay, where a mistranslated term is unrecoverable without re-translating the whole page. Breadth is the second strength. Document handling covers PDF, ePub, HTML, TXT, DOCX, Markdown and subtitle files, with OCR for scanned PDFs and layout preservation, so the output still looks like the paper or contract you were reading. On video it delivers real-time bilingual subtitles across 100+ platforms including YouTube, Netflix and Prime Video, and it will translate videos that ship with no subtitles at all. Meeting translation runs on Zoom, Google Meet and Microsoft Teams. Image translation uses OCR plus inpainting to keep the visual layout, and comic/manga translation covers Japanese manga, Korean webtoons and Western comics. Engine choice is genuinely open: the vendor lists 20+ engines including DeepSeek, DeepL, OpenAI, Claude, Gemini, Microsoft Translate and Google Translate, with customizable AI terminology libraries for domain-specific vocabulary. You can compare providers on the same passage rather than trusting one vendor's output. Where it fits less well: it is a reading and subtitle tool, not an interpreter. It doesn't do real-time spoken conversation translation for in-person meetings. Teams that need centralized billing, seat management and admin controls won't find them here — this is a per-person productivity tool. It also requires configuration: engine selection and terminology libraries are the difference between good and mediocre output, and users who won't spend an hour tuning that will get less out of it. Image translation, manga translation and Privacy Mode are labeled beta by the vendor. Deployment is desktop browsers plus iOS and Android, so there's no self-hosted or offline option described. Privacy claims are unusually specific for the category — no retention of translated content, nothing used for training, encrypted transmission, and ISO 27001, ISO 27701, GDPR and APPI listed. If your work involves sensitive documents, that matters more than engine count.