Turn a transcript into Premiere Pro sequence markers

20 September 2026 · 5 min read

Captions and markers look like the same data and do opposite jobs. A caption track is for the viewer: it renders, it gets delivered, it needs to read well. A marker is for you: it never appears in the output, it sits on the timeline at the frame where a phrase was spoken, and it is searchable.

If you have ever scrubbed through an hour of interview footage trying to find the moment somebody said the good line, that is the problem markers solve. The transcript already knows where it is.

Why the file is a .xml and not an FCPXML

This trips people up, because both formats descend from Final Cut. FCPXML is the modern Final Cut Pro format, and Premiere cannot read it at all. What Premiere reads is xmeml, the interchange format from Final Cut Pro 7, with some Adobe-specific tags layered on. It takes the plain .xml extension.

So the two files are not versions of each other, and picking the wrong one does not produce a degraded import, it produces no import. Export the FCPXML for Final Cut Pro or DaVinci Resolve, and the marker XML for Premiere.

Frame rate matters more here than it looks

Inside an xmeml file a marker is not a timestamp, it is a frame number, counted against the sequence's own timebase. There is no seconds value anywhere for Premiere to fall back on. Export at 25 fps and import into a 23.976 sequence and every marker is in the wrong place, increasingly so the further into the timeline you get.

Set the frame rate picker to match the sequence before exporting. If you uploaded a video file, the rate is usually detected from the file itself and preselected for you.

What lands in the Markers panel

• One marker per caption line, starting on the frame the first word begins and ending on the frame the last word finishes.
• The marker name is the start of the line, because that is the column the Markers panel shows.
• The marker comment holds the full line, which is what marker search actually matches against. So a phrase halfway through a long caption is still findable even though the name column truncates it.

Using it alongside captions

These are not alternatives. Exporting the SRT and the marker XML from the same transcription costs nothing extra and they are built from identical word timings, so they cannot disagree about a frame. Import the SRT onto a caption track for the deliverable, and the marker XML for editing against.