Notes between the captions

A transcript is a caption and its gloss. It has nowhere to say the other thing — that this idiom is the one from the first lesson, that the speaker has just changed register, that the next four lines are a song and not speech. A note is that somewhere.

1. Writing one#

Between any two captions — and before the first, and after the last — is a seam, with a faint + in it that brightens when the pointer is near. Press it (write a note here) and a new note is made, anchored to that place, and opens straight in the editor, in a sheet over the page. It starts with a title, A note (or A note 2, and so on, so that each of a video’s notes has a name of its own), and a line of text to write over.

It is a studio document — not like one: the same editor writes it, the same renderer draws it, and every mark of the dialect means what it means in the studio, pictures, recordings and exercises included (the section Studio explains the editor and the section The Markdown dialect what it takes). A note’s title is its name among this video’s notes: two notes of the same video link to each other by it — [as I said](doc:On the greeting) — the name must be unique there, and renaming a note follows every link to it, exactly as in the studio’s own library, but within this video’s notes alone.

2. Reading one#

A seam with a note in it carries a small mark: ✎ and the note’s title. Click it and the note opens over the page — never inside the transcript, because the transcript is the point, and a page with an essay between two captions is not a transcript any more. The video pauses.

The sheet’s bar has the title, edit (the note in the editor), read (back to the note as it renders) and ✕. Esc, the ✕, or a click on the darkened page around it closes it; the marks are then drawn again, in case the title changed.

It opens at once. What the sheet shows is the note on the studio’s own sheet — the same face, colours and marks — with no editor and no scripts behind it, and the player fetches the notes of the captions around you before you ask for one, so a mark opens instantly even over a slow tunnel. Open in the studio, in the sheet’s bar, is there when the full page is wanted; a note that holds an exercise opens that full page straight away, so nothing that cannot be answered is ever put in front of you.

They come with the video onto a phone. Keep on this phone lists them as one line of its own — its notes · 12 · about 96 kB — ticked to begin with, and unticked with a tap where you would rather not have them. Kept, the marks are in the seams of the transcript and the notes open with the computer away, in the studio’s own face and colours; one that holds an exercise can be answered there as it can at the desk. The pictures in them come along; their recordings are a line of their own beside the film, so nothing heavy is kept by surprise. A note written at the desk afterwards shows its mark as soon as the two can reach each other, and the video says it is out of date so the note itself is one tap away (Browser and Mobile).

Rest the pointer on a mark first, and it says what the note is. After a third of a second — so that sweeping across a seam on the way to a phrase does not light up every note it passes — a small card opens by the mark: the note’s title, the start of what it says (about 280 characters, the markup taken out), and click to open. The keyboard gets the same card: Tab to a mark and wait. It goes when the pointer leaves the mark, on Esc, on a click and on any scroll, and clicks pass straight through it. On a touch screen there is no card: a tap opens the note.

The card is letters and nothing else. A note is a file that travels — inside a video’s download, written in somebody else’s toolbox — and a card that opens because a pointer happened to cross a seam must never do more than show text. Nothing in a note is run or drawn there; clicking the mark, which opens the note itself, is a choice.

3. Where a note says it sits#

In its own front matter, one line:

---title: On the greetingtarget: faanchor: before cap 12---

before or after, then cap (a caption), then the caption’s start in seconds — what transcript.txt is keyed by, which is why a note stays put when the phrases of a caption are edited, cut or joined. The seam between the captions at 6 and 12 seconds is both after cap 6 and before cap 12; a note made with its + says the second, and one made after the last caption says after cap and that caption’s start.

Moving a note is editing that line, in the note’s editor or in the file: nothing else has to be told, because the file is the truth here as everywhere in Parseh.

A note whose anchor names a caption that is not there — a start that was moved, a line mistyped — is not dropped and is not an error. It is shown at the end of the transcript, its mark dashed, and its card says this note names a caption that is no longer in the video — open it and change its anchor line. That is where somebody will look for it.

4. Where notes live#

In the video’s own folder, never in the studio’s library, because a note is about this video:

youtube/videos/persian/<id>/markdown/persian/<note-id>/source.md

— the studio’s own layout (a folder per language, a folder per document, with its meta.json, its images/ and audio/), rooted in the video’s markdown/ folder.

Two things a note does not do:

It is never built.

A note is read on its page and nowhere else: there is no PDF of it and no .tex, and the server refuses both — a note beside a book or a video is read on its page and nowhere else: it is never built to a PDF and never taken away as LaTeX. Copy it into the studio if you want either.

It is in no prompt.

Nothing Parseh hands a model mentions notes, and nothing should: a note is your own reading, which is exactly what a model has nothing to put in.

Notes travel with the video: the ⤓ download carries every note of it, pictures and recordings included (an .svg figure excepted: it stays on the machine it was drawn on), and bringing the zip back puts them back (Taking videos away).