Subtitle to Plain Transcript

Subtitle files are written for video players, not readers. An SRT file for a 30-minute interview contains several hundred cue blocks. Each block is a timestamp, 1–3 lines of text, and a blank line. The same sentence is often split across two or three cues because the reader has time to read it between the start and end times on screen. The result: a file where every paragraph break falls in the middle of a thought, timestamp lines interrupt every 2–5 seconds, and the text structure is dictated by reading speed rather than meaning. Paste that file into a document editor and it is nearly unusable as text without significant cleanup.

The AT USE Subtitle to Plain Transcript tool strips all of that structure and produces running prose. Drop in any subtitle file — SRT, WebVTT, ASS/SSA, or SBV — and the tool re-joins lines that were split across cue boundaries, removes timestamp entries and sequence numbers, and inserts paragraph breaks at gaps between cues longer than two seconds. The two-second threshold catches natural speech pauses — a speaker finishing a thought before moving on — without treating every brief pause as a paragraph end. The output is text you can paste into a CMS, hand to a translator, or index for search without touching up every line break.

Three things make this different from manually deleting timestamps in a text editor. First, it handles cue-split sentences: when a sentence ends partway through one cue and continues at the start of the next, the tool joins them into a single line rather than leaving a hard line break mid-sentence. Second, it deduplicates consecutive repeated lines — a common artifact in YouTube SBV exports, where the rolling caption re-states the previous line on the new cue. Third, it strips ASS/SSA override tags ({\b1}, {\i1}, positioning codes, color codes) that appear as literal text in a plain-text export if the file is just re-saved with a different extension.

Output formats

TXT output is unformatted plain text with double newlines between paragraphs. Speaker labels (when present in the source and speaker label detection is enabled) appear as "Speaker Name:" on the line before their paragraph. This format works for everything: CMS content editors, CAT translation tools, Word, email, Slack.

Markdown output wraps speaker labels in bold using **Speaker Name:** and preserves paragraph breaks as Markdown blank lines. Use Markdown when the output goes into a static site generator, GitHub README, Notion, or any CMS that renders Markdown natively — the formatted labels give the transcript structure without requiring manual formatting after export.

Periodic timestamp markers

The optional "Add timestamps every N minutes" setting inserts a [MM:SS] marker at the nearest paragraph boundary to each interval. A 45-minute webinar transcript with markers every 15 minutes gets four timestamps inserted at approximately [00:00], [15:00], [30:00], and [45:00]. These markers serve as navigation anchors in published transcripts and are exactly what podcast apps (Apple Podcasts, Spotify) display when a transcript has chapter-style navigation. The markers are placed at paragraph breaks, not mid-sentence, so the text reads naturally around them.

Privacy

All processing runs in your browser. Subtitle files are read locally and never sent to any server. A 3-hour interview transcript in SRT format processes in under 2 seconds on a modern laptop, with no network request after the page loads.