Premiere Captions to Word

Three things Premiere can give you, and only one is this job

A finished sequence often leaves the edit in three shapes that people all call “the captions.” Burn-in draws the words into the picture, so there is no text file to convert — you would be asking Word to read pixels. Embedded or broadcast sidecars (SCC, MCC, STL, and some XML) are timed caption streams for playout systems. They do not use SubRip’s cue number plus an --> arrow, so this parser will not turn them into paragraphs. The third shape is a SubRip sidecar: plain cues, one language stream, clocks that are supposed to match the sequence. That is the file to drop here.

Adobe’s export notes say an SRT sidecar supports one stream per file, even when the sequence has several caption tracks (608, 708, subtitle, and so on). If the client asked for English and Spanish, you export twice. Mixing those tracks into one SRT is a Premiere choice you make before you ever open this site, not something the converter can guess from a video.

Export the sidecar from the sequence

With the sequence active, choose File → Export → Media. In the export dialog, open Captions and set the export option to Create Sidecar File. Set the file format to SubRip Subtitle Format (.srt). Turn SRT styling off. Export. You get the video (or you can cancel the picture encode if you only needed the sidecar — the caption file is the point of this page) plus an .srt next to it.

Styling off matters here in a way it does not matter on YouTube’s SRT download. Premiere can write the caption’s font and color into the cue as tags such as <font color="#ffffff"> and <b>. Players that expect a plain SRT often show those tags on screen. This converter treats them the same way: they are characters in the cue text, so they become characters in the Word paragraph. The homepage strips WebVTT voice tags and ASS override codes. It does not strip Premiere’s HTML-style caption styling. If the preview shows a font tag, do not “clean it up” by hand across a one-hour show. Re-export with styling off and drop the new file.

The Captions panel’s three-dot Export to SRT file is the path that skips the styling checkbox. Editors use it because it is one click and does not render the video. The file is still named .srt, and it still has arrows, so it will parse — and the tags come with it. Use that menu only when you have already confirmed a short sample has no style markup. Otherwise use Export Media.

Check that cue 1 is the sequence, not zero

There is a second export that looks identical in a text editor and is wrong for a client document. Selecting the caption item in the Project panel and choosing File → Export → Captions can write the file from the caption clip’s own start. Cue 1 becomes 00:00:00,000 even when those words appear twenty minutes into the sequence. Premiere ignores incoming caption timecode when it places a caption clip on a timeline. The sidecar you want is the one written on the way out, from Export Media, so the ranges match the edit.

Before you download a .docx, open the SRT and compare three cues to the timeline: the first spoken line, one in the middle, and the last. A constant gap on all three is an offset you should repair in Premiere (or in a subtitle editor) before Word. A document that prints the wrong range is not a QC file. The fix SRT timing page is that diagnosis. Do not type new clocks into the .docx and call the captions fixed — nothing in Word writes back to the sidecar.

The transcript tab is not the caption track

Premiere’s Text panel has a transcript (speech to text, text-based editing) and a captions tab. A transcript can exist for the whole sequence while the timeline still has no caption track. Exporting “the text” from the transcript tab is often a .txt or a transcript document, not a SubRip file. Pasted here, a dump with no --> lines produces no cues. Create captions from the transcript inside Premiere if you want timed captions, then export the sidecar as above.

Speaker names follow the same rule as any other SRT on this site. If the cue starts with [Alex] or Alex:, merge can keep that person as a bold label. Auto-generated Premiere captions often have no speaker at all. Merge will then join nearby lines into one voice, which is what you want for a single narrator and the wrong default for an unlabeled two-person interview. For interviews, add the labels in the SRT first, or follow interview SRT to Word. This page does not detect speakers from the audio — there is no audio on the way in.

Download two Word files and leave the SRT alone

Drop the sidecar with Show timestamps on and Merge same speaker on, unless the cue itself holds two languages (then merge off, and read the bilingual guide). The preview should be paragraphs with ranges, not cue numbers and not <font. Download premiere-timed.docx for anyone who will quote a minute. Then use the reading-copy link, confirm the timestamp box is empty, and save premiere-reading.docx for the script or the blog draft.

Keep the .srt next to both documents. Premiere imports SubRip back onto a caption track. It does not import a .docx. Wording that must appear on the video gets edited in the caption track or the SRT. Wording that is only for a reader gets edited in Word. If the same video is also going to YouTube, the Studio caption download is a different file from this sidecar — use YouTube captions to Word when the source is Studio, not when the source is the sequence. CapCut’s phone-width lines are a third workflow.

One SRT still converts with no account, free forever. Batch ZIP of many sequences is Pro (batch ZIP + header/footer) on the converter, not a fee for reading this page, and not an Adobe subscription.

Related workflows

• Fix SRT Timing Before Word
• YouTube Captions to Word
• CapCut Subtitles to Word
• All subtitle-to-Word guides
• SRT to Word converter — same tool the buttons above open, with these toggles already set.