Which sites NoaLingua puts dual subtitles on

Why the platform list is shorter here than elsewhere

Two content scripts do this work. One handles text selection and runs on any page. The other looks for a video element with a caption track it can read, and that second one is the whole story of platform support: where a site exposes such a track, the subtitle box appears; where it does not, nothing does.

Beyond that, YouTube and Netflix have adapters written specifically for them, which is what makes sentence rebuilding, replay on real boundaries and — on YouTube — look-ahead translation and transcript export possible. A generic caption track gives none of that even when it is readable.

Several competing products list a dozen platforms or more, and that is a real advantage of theirs rather than marketing. If most of your watching happens on Disney+ or Prime Video, they are the better answer, and the comparison pages say so in as many words.

The part that works everywhere

Subtitles are one of four surfaces. On any web page you can select text for the same translation card the video gives, hold Alt and click a paragraph to place a translation underneath it without replacing anything, or highlight every word your deck has never met. An article in your target language is fully supported on any site.

PDFs are the fourth. A document can be read with paragraphs aligned side by side, or in its original layout with every word still clickable, and either way the words you save join the same schedule as the ones saved from a video. None of that depends on a platform choosing to expose captions.

What is never possible, on any site

Nothing here bypasses DRM, geo-restriction or a subscription. The extension reads what the site has already displayed to you, in the region and on the plan you already have. A title you cannot watch stays a title you cannot watch, and a language track that is not offered where you are does not become available.

It also does not create subtitles where none exist, unless you deliberately turn on speech recognition — and then it labels the result a machine guess, because that is what it is. There is no mobile app and no Firefox or Safari build, so none of this applies on a phone or a television.