NEW Subtitle OCR plugin released

[Marc Brooks (IDisposable)]

I've just released Subtitle OCR, a new Jellyfin plugin that turns embedded image subtitles into searchable, editable text files.



What it does

• Files: All Movies and Episodes in your Jellyfin Libraries
  
• Inputs: VobSub (DVD) and PGS (Blu-ray) image subtitles from any container ffprobe supports.
  
• Outputs: SRT, ASS, or Auto (ASS only when positioning requires it, SRT otherwise).
  
• Uses pure managed .NET code (nOCR pattern matching) — no Tesseract, no native dependencies.
  
• Fully compatible with Subtitle Edit .nocr databases (drop in ones you've trained or corrected).
  
• Smart features: language detection for untagged tracks, conservative text fixes (l/I, contractions, etc.), optional Hunspell spell-check, protected metadata words, and safe naming ({name}.{lang}.OCR.srt) so it never overwrites your existing subtitles.
  
  

Tested on a 14,000+ item library running Jellyfin 10.11.

Installation (Recommended)

1. Dashboard → Plugins → Repositories → Add:
   Code:

   https://raw.githubusercontent.com/IDisposable/jellyfin-plugin-subtitleocr/main/manifest.json
2. Go to Catalog → Subtitles category → Install Subtitle OCR.
   
3. Restart Jellyfin.
   

Beta/pre-release channel available with the -beta manifest. Manual .zip install also supported.

GitHub: https://github.com/IDisposable/jellyfin-...ubtitleocr

Full README with features, config options, architecture, and known limitations is there.
Feedback, bug reports, and contributions are very welcome!

Especially interested in testing with non-Latin languages (only Latin database is bundled right now).
Let me know how it works on your library!