How to convert VobSub DVD subtitles to SRT on Mac

Find the index and the pictures

Suppose you’re restoring the subtitles for a museum film your team produced. The handoff folder contains Pressroom.idx, Pressroom.sub and a reference movie. Open a copy of the IDX in a plain-text editor. A VobSub index commonly begins with a VobSub version line. Farther down, a language entry such as “id: en, index: 0” is followed by timestamps and file positions. Each position points into the companion SUB file. The words you see on screen are stored there as pictures.

Keep both files together with their matching base name. Work on copies and keep the original reference movie with the archive. If you received only the IDX, ask for its matching SUB. Reading the timestamps will tell you when pictures appear, but cannot recover the letters drawn in those pictures. A .sub extension alone is ambiguous. If your file opens as readable caption sentences, identify its text format before attempting this workflow.

Choose the track you mean to recover

Play the reference cut with its original subtitle track selected. Write down the language, an opening line, a distinctive later line and their times. If the menu offers both English dialogue and English captions for deaf and hard-of-hearing viewers, find a passage with an offscreen sound. The presence of a caption for that sound can distinguish two tracks whose opening dialogue is identical.

Ask the archive owner whether an approved SRT for this cut already exists. Compare those reference lines and times if one turns up. Using the approved file can save an entire proofread. Otherwise use optical character recognition (OCR) to read the letter shapes in the pictures. Transcribing the soundtrack would create a different editorial record. A narrator may pronounce an abbreviation in full, and a door closing may have a carefully chosen caption. Neither tells you exactly what the original subtitler wrote.

If the archive supplied an MKV, check the track list before selecting a subtitle row for conversion. Matroska stores each stream from a multilingual VobSub set in a separate track. Write down which track produced your reference line and check that same line in the exported SRT. Converting one English track does not recover the other languages in the container.

Convert the selected VobSub track in Subler

Subler for Mac lists raw VobSub IDX as a supported input. Make a new document and add the IDX from your working folder, with the SUB beside it. If you have a permission-cleared movie containing the VobSub track instead, add that movie. In the import list, identify the subtitle row for the language you checked. Keep track of that row if several languages are present.

Follow Subler’s Subtitles Guide and select the Tx3g conversion action for the VobSub row. Tx3g is timed text that can be stored in an MP4. If Tx3g is missing, check Advanced preferences for the “Convert VobSub to Tx3g” option described in Subler’s guide, then reopen the import list. Passing the bitmap track through would keep it as pictures. For a language that needs additional recognition data, Subler’s FAQ points to the OCR tab in Settings. Choose data for the writing on screen, even if the film’s spoken language is different.

Save the converted document under a new name. Select its resulting text subtitle track and export it as an SRT. Give the file a working name such as Pressroom-en-ocr.srt. Open it in a text editor and look for numbered cues, time ranges and readable words. An MP4 that plays with captions is only an intermediate result here. The file your editor needs is the separate, editable SRT.

Proofread against the original pictures

Put the original bitmap track on one side and the draft SRT on the other. Read each bitmap before looking at the recognized text. A fluent sentence can conceal an error as easily as a string of nonsense. Names, accents, identifiers and short standalone captions deserve a deliberate second pass. A spelling checker has little basis for choosing between two plausible shelf marks.

Consider this invented example from the museum film. At twelve seconds the caption names a tray in the collection. At sixteen seconds the press stops. Compare what the original pictures show with one possible OCR draft.

12.000–15.000
Image: Inès, fetch tray I-07.
Draft: Ines, fetch tray 1-07.

16.000–18.000
Image: [press stops]
Draft: (cue missing)

Restore the missing cue

The draft has three faults. Restore the accent in Inès. Check the tray’s identifier against the bitmap and, if available, the museum’s inventory. Here the first character is a capital I. A digit 1 would direct someone to a different tray. Search for other instances of this identifier individually. A global replacement of 1 could change an unrelated measurement.

Now play through sixteen seconds with only the original captions visible. If the sound caption appears, note when it comes on and clears. Insert its own timed cue in a working SRT using those boundaries. Adding “[press stops]” to the previous line would make it appear while Inès is still receiving the instruction. Compare adjacent cues after inserting it. The following example shows the two recovered cues in a standalone excerpt.

1
00:00:12,000 --> 00:00:15,000
Inès, fetch tray I-07.

2
00:00:16,000 --> 00:00:18,000
[press stops]

Return a file the next editor can use

Review the full sequence against the original bitmap track. Work through the source pictures in order so an omitted cue attracts attention. Then reverse the comparison and check that every SRT cue corresponds to a picture. Note any unreadable word with its time and send it to the film owner for a decision. Keep the untouched OCR draft alongside the corrected version so another reviewer can see what changed.

For a playback pass, import the corrected SRT into Polyglot and attach the matching reference video. Polyglot can review, edit and export the timed text after the external OCR step. Compare the recovered opening line, the shelf mark, the sound caption and the last line against the approved cut. Disable the old track when previewing the new one. Two captions drawn on top of each other make that comparison hard to trust.

Deliver the reviewed SRT with the cut identifier and language in its filename. Open the matching movie in the recipient’s player, load the delivered SRT explicitly and check the accent, identifier and sound cue once more. A plain SRT carries the words and their times. Retain the IDX/SUB pair and reference movie for anyone who needs to see the original type, colour or placement.

Keep exploring.