Step 2 · Text· 5 min read

How to keep captions in sync after cutting your edit in Premiere Pro

In short

After an AI transcription, Subkick Caption anchors every word to the moment it is spoken in your source clip. If you then cut, move, trim or delete clips in your Premiere Pro timeline, the words realign on their own when you come back to the panel, before each generation and after an Autocut; the 🔗 button (Re-sync words with the edit) in the Text step forces it at any time. This works for transcribed words only: imported or typed captions keep their timecodes.

Step by step

Transcribe first — Add your text with + Add text › ✨ AI transcription: every word is anchored to the clip it comes from.
Edit in Premiere Pro — Cut, move, trim or delete clips in your timeline as usual.
Come back to the panel — Click back into the Subkick Caption panel: the words realign and a message says how many were re-synced.
Force it if needed — In the Text step, click 🔗 above the timeline to re-sync the words with the edit right away.
Update your captions — Generate again with ⚡ Update: only the blocks that changed are redone on the timeline.

What are anchored captions?

When you transcribe with AI transcription, Subkick Caption remembers more than the time of each word in the sequence: it also records which clip the word comes from and where it is in that clip. That’s the anchor.

When your edit changes, the panel reads the clips again and puts each word back where its audio now sits. You can keep cutting after transcribing without re-transcribing and without retiming by hand.

Why it matters: captions usually come at the very end, because any late change breaks the timing. With anchored words, you can caption early, show a first version to a client, then keep cutting — the captions follow the edit, and the transcription isn’t paid twice.

When do the words re-sync?

When	What happens
You come back to the panel after editing in Premiere Pro	The panel checks whether the clips of the transcribed tracks changed. If they did, the words realign and a message says so.
Before each generation	The check runs first, so generation never uses outdated times.
After an Autocut	The words follow the cuts right away.
You click 🔗	In the Text step, above the timeline (Re-sync words with the edit (timecodes follow your cuts)): forces a re-sync even if nothing seems to have changed.

The result is shown in a message, for example 🔗 42 words re-synced to the edit., sometimes followed by details: 3 removed (their audio is no longer on the timeline)., 2 blocks split at a cut. If everything is already in place: 🔗 Words are already in sync.

What happens to the captions for each kind of edit?

You…	The captions…
Move a clip	Follow it. The order of the words and the absence of overlaps are kept.
Delete part of a clip, or trim it	Lose the words whose audio is no longer on the timeline.
Cut in the middle of a block, so that a cut or a gap ends up between two of its words	The block is split at that point: each part stays with its own audio.
Change the speed of a clip (constant speed, or played backwards)	Are placed exactly at the new times.
Use time remapping with keyframes	Are placed approximately; the message says how many: N words approximate (time remapping).
Stack two clips so their voices overlap	The first word stays on screen; the next one starts when it ends.
Place the voice clip inside another sequence, or in a multicam clip	Are still followed, through the sequence that contains it.

Words you retimed by hand keep your timing: your edit becomes the new reference. Words you added by hand keep their spacing with the word before them.

When is the re-sync refused?

🔗 Re-sync refused: N of M words can’t find their audio anymore (media replaced?). Re-transcribe if needed. More than half of the words lost their source — typically after replacing the media. Nothing is changed. If the audio really is different, transcribe again.
🔗 No anchored words: transcribe first. The text comes from an import or from typing.
⚠ This subtitle session belongs to another sequence: open it in Premiere to resync. Open the sequence these captions belong to.
Busy: wait for the current operation to finish. A generation, transcription or Autocut is running.

Which captions are anchored?

Only the words that come from AI transcription. Captions imported from an .srt / .vtt file or from the project, and text typed in ✍ Write, keep their timecodes and don’t follow later edits — see Import an SRT file or your Premiere captions.

Some actions rebuild the whole text and drop the anchors: a new import, new typed text, ✂ Re-split and 🗑 Delete all text. Splitting (×), merging (⇄) and fixing words keep them. If 🔗 says there are no anchored words any more, transcribe again.

Tip: to change how lines are cut, set Characters per block in Style › 🔧 Layout › ✂ Subtitle splitting before transcribing, rather than re-splitting afterwards.

Which tracks are watched?

The panel looks for the audio on the tracks you transcribed:

with All non-muted tracks, it watches every audio track of the sequence — muting a track later doesn’t remove its words;
with one or more tracks ticked, it watches only those: keep the voice on these tracks, otherwise its words can’t find their audio.

How to undo a re-sync

A re-sync is a single step in the panel’s history: Ctrl+Z / Cmd+Z, with the panel in focus, brings the previous text and times back. It doesn’t undo your edit in Premiere Pro.

A smooth workflow

Transcribe as soon as your rough cut is ready.
Style and lay out your captions — see Layout and word placement.
Keep editing in Premiere Pro: cut, tighten, reorder.
Come back to the panel, check the message, then ⚡ Update to refresh only the blocks that changed — see Generate on the timeline.

Frequently asked questions

Do captions follow my cuts if I re-edit the video in Premiere Pro?

With Subkick Caption, yes for transcribed captions. Each word is anchored to its moment in the source clip, so when you cut, move or trim clips after transcribing, the words realign on their own when you come back to the panel. Generate again with ⚡ Update to refresh the captions on the timeline.

What does the 🔗 button do in Subkick Caption?

The 🔗 button, above the timeline of the Text step, re-syncs the words with your edit on demand, even if the panel thinks nothing has changed. Subkick Caption then reports how many words were re-synced, removed or split at a cut.

Why does Subkick Caption say “Re-sync refused”?

More than half of the words can no longer find their audio, which usually means the media was replaced. To protect your text, Subkick Caption changes nothing; transcribe again if the audio really changed.

Do imported SRT captions follow my edits?

No. In Subkick Caption, only words from AI transcription are anchored to the clips. Captions imported from an .srt file or from the project, and typed captions, keep their timecodes: finish your edit first, or retime them in the Text step.

Does re-syncing the captions use transcription credits?

No. Subkick Caption re-syncs the words by reading your timeline, on your computer: nothing is sent and no credits are used, however many times you re-edit.