Convert Blu-ray SUP subtitles to SRT
Read the text out of PGS bitmap subtitles and save them as SRT. The OCR runs on your device — no upload, no queue, no account.
Blu-ray subtitles are not text. Each line is a picture: a run-length-encoded bitmap with its own colour palette, timed to appear and disappear. That is why a .sup file survives every converter unchanged, why you cannot search it, restyle it or feed it to a translator, and why the tools that do convert them almost all ask you to upload the file and wait in a queue — optical character recognition is the expensive part, so it is the part they charge for. Your browser is perfectly capable of doing it. Subtitle Doctor decodes the bitmaps and their palettes, runs recognition locally, and shows each original picture beside the text it read so you can check the ones that look wrong. Nothing is uploaded, including the recognition model, which is served from this site rather than a third party.
Worth knowing
- A PGS file contains no duration field anywhere. A subtitle ends when a later composition segment names no objects at all, which is the format’s way of saying "clear the screen" — that empty composition is what supplies every end time.
- Each original bitmap is shown next to the text recognised from it, because OCR on subtitle images is never perfect and the fastest way to check a suspicious line is to look at the picture it came from.
- Bitmaps are inverted and composited onto white before recognition, since Tesseract expects dark text on paper while subtitles are light text over transparency. Getting that backwards fuses the background into the letters.
- Images are upscaled before recognition. Tesseract is markedly more accurate on glyphs around 30 to 40 pixels tall, and a 720p disc gives roughly 20.
- The recognition model is about 4 MB and is served from this site, not from a content delivery network, so this page makes no third-party requests like every other page here.
- Results are handed straight to the editor, because an OCR pass produces a draft rather than a finished file and the wrong letters are much easier to fix against the video.
FAQ
Why can’t I just convert a .sup like any other subtitle file?
Because there is no text in it to convert. SRT, VTT and ASS all store characters; PGS stores pictures of characters, drawn at authoring time with a specific font and palette. Turning one into the other means recognising the letters in an image, which is a fundamentally different and much less certain operation than rewriting timestamps.
How accurate is the recognition?
Good on clean, standard fonts and noticeably worse on italics, unusual typefaces and heavily anti-aliased small text. Expect to correct a handful of lines. Common OCR confusions — a capital I read as a pipe, a zero for a letter O — are exactly what the fix-common-errors tool is built to repair, so running that afterwards clears most of what is left.
Does the file get uploaded for processing?
No. The bitmaps are decoded in your browser and recognition runs there too, using a model served from this site. This is the whole reason the tool exists in this form: OCR is the step other subtitle sites run on their own servers and meter, and there is no technical reason it needs one.
What about DVD subtitles, VobSub or .idx and .sub files?
Not yet — those use a different container and a different bitmap encoding, wrapped in MPEG program stream packets. This tool reads PGS, which is the Blu-ray format found in .sup files and in S_HDMV/PGS tracks inside MKV. If you extracted an image track from an MKV here and it came out as .sup, this is the tool for it.