Split and Merge Subtitle Timing
When Whisper groups syllables wrong or a word boundary is off, split or merge to fix it. Split divides a word at a precise timestamp. Merge combines adjacent words into one.
How It Works
Splitting a Word
-
Find the word in the timeline: look at the word chips row below the main caption track. Each chip represents one word with its time span.
-
Right-click at the split point: right-click anywhere on a word chip in the timeline row. The split happens at the exact timestamp where you clicked. This gives you frame-accurate control over where the boundary falls.
- Or use the scissors button: for chips wide enough (52px or more), a scissors icon appears. Click it to split at the midpoint of the word. Useful when you just want a quick 50/50 split.
- Edit the new word: after splitting, the second word (wordB) automatically enters edit mode. Type the correct text for the second half. The first word (wordA) keeps the original text.
Merging Words
-
Select adjacent words: hold Cmd and click two or more adjacent words in the transcript sidebar or the timeline row. Selected words show a blue ring.
-
Press Enter to merge: the selected words combine into a single word with the merged text and the combined time span.
Key Details
Splits are non-destructive. The second word created by a split is marked as _edited and preserves the original text. Revert is available via right-click context menu.
Merge requires contiguous words. You can only merge words that are adjacent: no gaps allowed. If you select non-contiguous words, the merge won’t trigger.
Timestamps stay accurate. Split creates wordA (start to split-time) and wordB (split-time to original end). Merge creates one word spanning the full range. No gaps or overlaps are introduced.
What’s Next
- Edit Words in the Transcript: fix text with confidence-guided editing
- Per-Phrase Caption Positioning: move captions to different locations per phrase
- Find and Replace Across Transcript: batch-correct repeated misspellings
Keep the edit on your Mac.
Choose a license to record, clean, caption, and export without moving private footage into a browser editor.