NebulaCafe

ChatGPT plugin and Claude connector

Subtitle Repair

Check subtitle structure. Normalize the file. Keep the words.

Use Subtitle Repair in ChatGPT with an existing plain-text UTF-8 SRT or WebVTT file. Inspect numbering, ordering and overlaps before making changes. Export normalized SRT, WebVTT and a JSON audit report, with an optional explicit timing offset and stable sort.

Using Claude? Read the connector docs.

Watch it in ChatGPT

Inspection, export, download and reopening a saved file. Idle waits are removed; remaining footage plays at 1.25×. This demonstration uses synthetic subtitle text.

What it supports

  • Audit up to 2,000 cues in a plain-text UTF-8 file up to 512 KB.
  • Normalize numbering and formatting without rewriting cue text or changing cue duration.
  • Apply an explicit millisecond offset and optionally sort by cue timing. Intentional overlaps are retained.
  • Download SRT, WebVTT and the exact JSON report, or reopen an existing export.

Subtitle Repair does not transcribe, translate or align captions to speech. It rejects invalid timings, unsupported styling and offsets outside the supported timeline instead of guessing repairs. It does not certify platform or accessibility compliance. A connected NebulaCafe account is required.

Your subtitle files

NebulaCafe processes the subtitle text you supply. Inspection does not save a new export. Exporting saves SRT, WebVTT and JSON files in Vercel Blob storage so you can download and reopen them. Anyone holding an export link can access that file. Avoid including confidential information in exported subtitles.