NoaLingua for Chrome
Add to Chrome — Free Add to Chrome

Feature

Video difficulty, measured against your own deck

Every word spoken, compared with the words you know — before you press play

The transcript panel compares every word spoken in the video against your own deck and estimates how hard it is for you specifically. It lists the words you have never met, ranked by how often they occur in that video, and names the grammatical constructions it uses, with an example of each.

Fit

Who Video difficulty is for

Choose it for the workflow it actually supports, not for a promise the product does not make.

Best for

  • Learners choosing what to watch next and wanting a filter better than a guess at their own level.
  • People who prefer to learn three words before a video rather than look up thirty during it.

Not the right fit for

  • Netflix, where no whole subtitle file exists to analyse before anything has appeared on screen.
  • Brand new decks: with almost nothing saved or marked known, every video is estimated as hard.

How to use it

How to use Video difficulty in NoaLingua

  1. 1

    Open the transcript panel

    Quick menu → Show / hide transcript. It docks into YouTube's own sidebar above the recommendations, and floats over the player in full screen, where no sidebar exists to dock into.

  2. 2

    Read the difficulty estimate

    It is computed across every word spoken in the video, compared with what is in your deck. If a video is noticeably hard, the same estimate is also shown once over the player — so you know before pressing play rather than after ten minutes.

  3. 3

    Look at the new words first

    The words you have never met, ordered by how many times they occur in this video. Each is a chip that opens the full word card. Three of them learnt in advance can change a video from exhausting to followable.

  4. 4

    Check the constructions

    The grammatical constructions the video uses, each with an explanation, a count of how often it occurs, and an example from the video itself.

Mechanism

Why NoaLingua behaves this way

Why grading against your deck beats a fixed level

A graded library sorts videos into beginner, intermediate and advanced against an average learner. That is useful and it is not about you: it cannot know that you have watched sixty hours of cooking videos and none about politics, which makes one "intermediate" video easy and another impossible.

Comparing the transcript against your own saved and known words produces a number about this video and this learner. It also improves as you use it, without anybody grading anything.

Where the whole transcript comes from

On YouTube the entire caption track is fetched once per video, which is what makes this possible at all: every word spoken can be counted, not only the ones that have already gone past. The same fetch is what gives replay its exact sentence boundaries and makes a bilingual PDF of the whole video possible.

The panel is also a transcript: the whole video as sentences with translations, a search field, and a Read as text mode that turns it into flowing prose with every word clickable. Clicking any sentence seeks the video there.

The same idea applied across everything you watch

Settings → Study carries the version that spans videos: "Learn these next" ranks twenty words by how many of your own videos they occur in, and states what share of the unknown words in your material they would cover.

What makes both possible is a deliberately small record — which words appeared in which video, and how many times. Numbers only. No transcript text, nothing about what was said, and nothing that leaves your machine.

Common questions about NoaLingua

Do I have to open the panel to see it?

No. When a video is noticeably hard, the estimate appears once over the player. It closes on a click anywhere and disappears by itself after twenty seconds.

Does marking words known change the estimate?

Yes, and that is a reason to use the Known status rather than ignoring it. A word marked known counts as known everywhere: the estimate, the highlighting, the new-word lists and the review queue.

Can I search the transcript?

Yes — the panel has a search field with next and previous buttons to jump between matches, and clicking any sentence seeks the video to it.

What if the captions have no punctuation?

Where the punctuation model has been downloaded and more than half the transcript lacks punctuation, it is restored — but only if the model returns exactly the same words. Change one word and the original is kept.