Edit transcripts

Fix typos, split, merge, find and replace, and word timings stay accurate when you don't touch them

Transcription gets you 95% of the way there. The last 5% is names, jargon, and the words your speaker swallowed. The transcript editor is where you fix that without breaking the timing.

Open the editor

Four ways in:

• Right-click a clip → "Edit Transcript…"
• Select a clip and press the keyboard shortcut bound in your settings
• Double-click a row's segment in a Paper Edit
• Open from a Mosaic timeline's "Edit Captions" action (different mode, captions live in memory until exported)

Two view modes

Toggle modes with the picker at the top of the Captions tab.

Segments mode is the default. One row per spoken segment, with editable text fields, start and end times, and per-row controls. This is where you fix words.

Script mode is a flat read-only transcript view. Drag-select word ranges, right-click the selection, commit them to a paper edit. This is where you mine for usable lines.

How edits keep timing accurate

Click a word, the player seeks to that word's start time. Edit text inline in any segment row. The thing worth understanding: when you change text, ReelChest matches old words to new ones and any word you didn't touch keeps its original timestamp. New words you added get evenly interpolated between the words around them.

Fixing a misheard name leaves every other timing intact. It's not magic, it's diff matching, but the result is clean captions without re-aligning by hand.

The four shortcuts that matter

• ⌘\ splits a selected segment at its word midpoint. Use this when one segment runs too long for a caption line.
• ⌘J merges the selected segment with the next one. Use this when the segmenter broke mid-sentence.
• ⌘F opens find. Substring, case-insensitive, across all segments.
• ⌥⌘F opens find and replace. Same matching rules. Useful for renaming a guest you misspelled the same way 40 times.

Other essentials: Delete removes selected segments. ⌘A selects all. ⌘Z undoes one step at a time. ⌘S writes to disk.

The word-per-line and line-count sliders at the bottom rebreak segments using your chosen layout. Words per line runs 1 to 12 (default 6); lines per caption runs 1 to 3 (default 1). Tighten or loosen to taste.

A working tip: don't try to fix every typo on first pass. Run find and replace on the four or five terms you know it got wrong, then ship the rough cut. Tighten the rest in the captions pass when you can see how the words actually land on screen.