LLL Article Transcript Export / 動画の文字起こしを記事用に整形

Questions

Which files work? WebVTT text whose first line is WEBVTT, up to 48 KB and 1,500 cues. SRT, YouTube links and audio or video files are not read.

What does the output look like? [mm:ss] Speaker: text, with [h:mm:ss] for every line when a cue starts after one hour. Markup tags are removed, HTML entities are decoded and voice names (<v Name>) become speaker labels. Paragraph layout joins cues from the same speaker and starts a new paragraph on a speaker change, a pause over 4 seconds, or at a sentence end after 30 seconds.

What is flagged? Cues with an invalid timing line, no text or no timing line are skipped. Cues that end before they start, overlap, run out of order or repeat the previous text are kept in the pasted order and listed by line number so you can check them. Nothing is reordered or guessed.

Is the transcript accurate? It is exactly the caption text you pasted. Check names and terms against the video before publishing.

Does it fetch or send anything? No. It has no network, content or storage access, and nothing is saved.

対象は？ 1行目がWEBVTTのテキスト（48 KB・1,500件まで）です。SRT・動画・音声・URLは対象外です。問題のある字幕は行番号で知らせ、並べ替えや推測はしません。

Security and privacy

Support and vulnerability reports: contact@lx3.cafe.

Requires a signed-in user who can edit drafts. The signed manifest requests no plugin capabilities and no network hosts. Only the pasted text and the layout choice are processed, in the EmDash sandbox. Nothing is read from your site, saved, fetched or sent by the plugin. Warnings name line numbers and do not repeat pasted text. Control, zero-width and bidirectional-override characters are removed from the output. Paste the result into your article as plain text and check it before publishing.

下書きを編集できるログイン済みの利用者が対象です。権限・通信先は要求しません。貼り付けた文字と出力の形だけを処理し、サイトの内容の読み取り・保存・取得・送信はしません。

0.1.0

• Turn pasted WebVTT into a plain time-stamped transcript, one line per cue or grouped by speaker and pause.
• Warn about invalid, reversed, overlapping, out-of-order, empty and repeated cues.

Install

Use EmDash 1.4 or newer with a sandbox runner. Search for LLL Article Transcript Export in the Registry and approve the install; it asks for no content, schema or network access. Open Article Transcript Export from the admin menu, paste the text of a WebVTT (.vtt) caption file, choose a layout and select Create transcript. Copy the result into your article as plain text.

Registryで LLL Article Transcript Export を検索して導入します。権限は求めません。管理画面のメニューから開き、WebVTT（.vtt）字幕ファイルの中身を貼り付け、出力の形を選んで 文字起こしを作成 を押します。結果は書式なしで記事に貼り付けてください。