SRT to Text

What SRT to text changes

A SubRip cue is built for a player. The player needs an index, a start clock, an end clock, and a short line that fits the frame. A person who wants to read, quote, or edit the words needs sentences. SRT to text is that second job. Cue numbers and arrow lines are dropped. Caption wraps become spaces inside a paragraph when neighboring cues join. Speaker labels that were written as [Maya] or Maya: stay as names on the paragraph.

The clocks are omitted from the reading copy because the checkbox above is already clear. They are still in the .srt you dropped. Nothing on this page rewrites that file. If the video still has to show the lines, keep the caption file beside the document. Word is where the sentences live. The player is where the clocks live.

People reach for a plain-text dump when the next tool wants a paste: a notes app, an email, a treatment, a translation brief that is about wording. Copy formatted puts the live preview on the clipboard as HTML and as plain text. Download .docx when the next person will open Word, Word Online, LibreOffice, or a Google Docs upload. Both follow the checkboxes. Neither one is a new subtitle file.

Steps

1. Open the .srt in a text editor. You want numbered cues and lines that contain -->. If the file is WebVTT, ASS, or lyrics, use the format page for that file first, then come back — or drop WebVTT and ASS straight onto the converter, which already reads them.
2. Delete cues you do not want as sentences. Music tags, applause tags, and a repeated sign-off at the tail will become paragraphs if you leave them in. This converter does not classify a line as junk.
3. Drop the file here. Confirm Show timestamps is off and Merge same speaker is on. The status line under the tool says so when the page preset applies.
4. Read the preview. Names should change when the voice changes. A wall of one speaker means the cues had no labels, which is expected for a single narrator and wrong for an unlabeled interview.
5. Download the .docx, or copy the preview. Save the original .srt under its own name. Download a second document with timestamps on only if a reader will ask which minute a sentence came from. That choice is written up on SRT to Word with timecode or without.

SRT to text versus SRT to TXT

The two pages are easy to mix up because both remove clocks. They do different work with the lines. This page joins them. SRT to TXT does not.

This is a three-cue file. The gap between Maya’s cues is two tenths of a second. The gap before Dev is six seconds.

1
00:00:01,000 --> 00:00:03,200
Maya: Thanks for joining.

2
00:00:03,400 --> 00:00:06,000
Maya: I have the notes up.

3
00:00:12,000 --> 00:00:15,000
Dev: Can we start with travel?

Run through the converter on this page, with the preset checkboxes, Maya’s two cues become one paragraph and Dev stays a second paragraph. There is no range on either paragraph. The words are “Thanks for joining. I have the notes up.” under Maya, then “Can we start with travel?” under Dev.

The same file through SRT to TXT is three blocks separated by blank lines: Maya: Thanks for joining., then Maya: I have the notes up., then Dev: Can we start with travel? No clocks, no cue numbers, and no join. Line breaks that lived inside a single cue stay inside that block. Italic and bold tags, if the cue had them, stay as characters in the .txt.

Use this page when the reader should meet paragraphs. Use SRT to TXT when a script, a diff, or a downstream tool wants the cue text and nothing else, still split where the captioner split it. Use the raw file in Word only as a quick look. You will still be reading subtitle syntax.

How the join decides a paragraph

Merge looks at two things: the speaker label and the gap. The label is the text of [Name], Name:, or a WebVTT <v Name> tag at the start of the cue. Alex and Alexander stay two speakers, because the label text is different. Two cues with no label are treated as the same speaker. That is why a single-narrator caption file becomes flowing paragraphs, and why an interview with no names becomes one voice.

The gap is the time from the end of one cue to the start of the next. About 2.5 seconds or less, with the same label, joins the text with a space. Extra whitespace collapses. A longer pause starts a new paragraph even when the name matches. In the sample above, Maya joins and Dev does not, because six seconds is longer than that gap. The pause is measured on the clocks in the file. A clock that is wrong will join the wrong lines. Repair the SRT before you trust the paragraphs — fix SRT timing is that repair, and this page does not shift a cue.

Turn merge off when each cue must stay its own block: bilingual pairs, a QC pass, overlapping speech you still need to see as two lines. The reading-transcript preset is the wrong preset for those files. Bilingual subtitles to Word explains the merge-off case. SRT without timestamps is the same timestamp checkbox, written as a warning against deleting clocks inside the caption file.

What the reading copy keeps

Dialogue text stays. A speaker label stays, shown in bold in the Word file when the cue had one. That is the SRT to text result: sentences a person can edit, with the cue furniture gone. Line breaks that existed only so a caption would fit a frame are not sacred. Once two cues join, they are one paragraph with single spaces. Punctuation is whatever the caption file already had. The join does not add a period, a comma, or a capital letter. If the captioner split a sentence across cues, you may need to fix the seam in Word.

Style tags such as <i>, <b>, and <u> stay as characters. They are not Word italics or bold. Positioning, karaoke, and ASS override codes are not an SRT feature; if you dropped an ASS file instead, those effects flatten to plain text on ASS to Word. WebVTT cue settings after the end time — align, line, position — are dropped. The words remain.

The document can also carry a short header line with the block count and the format name. Font, body size, and margins are under Setup: Calibri, Georgia, Times New Roman, Arial, or Cambria; 10, 11, 12, or 14 point; margins of 0.75, 1, or 1.25 inches. Header and footer text is the paid unlock, together with a folder of files as one ZIP. One file still converts without an account. The pricing page is the description of that unlock. This page does not change it.

A second file when someone will ask where

A reading transcript is a poor index. The minute a sentence was spoken is gone from the page, on purpose. Keep the .srt. If a producer, a lawyer, or you — later — will ask which minute, download again with Show timestamps checked. The range printed for a person is a short clock. Hours appear in that label only once the cue is past an hour. Milliseconds are not in that label. The full clock remains in the caption file.

Name the two documents for the reader. One is the prose. One is the check against the picture. Do not delete arrow lines inside the only .srt in order to manufacture a text file. That is how a caption delivery disappears. SRT to TXT and this page both leave the source file where you saved it.

Plain text, HTML, and PDF from the same preview

Copy formatted writes the preview as HTML and as plain text, then you paste into a document you already have. HTML download saves a standalone .html of that preview on your device. PDF opens the system print dialog so you can choose Save as PDF. All three follow the checkboxes you see. They are snapshots of the paragraphs. They are not subtitle files, and a player will not import them as cues.

Choose the .docx when comments, suggest mode, or a Drive upload matter. Choose the clipboard when the transcript is a section of a brief you already opened. Choose SRT to TXT when the requirement is literally a .txt and the cue boundaries must survive. The basic paste-and-download path for Word, including timestamps left on, stays on how to convert SRT to Word.

What this page will not do

SRT to text does not transcribe audio or video. If you do not already have an SRT, a VTT, an ASS file, or a Zoom-style transcript, there is nothing to convert. It does not translate. It does not detect speakers from a recording. It does not rebuild an SRT from the Word file. It does not shift, stretch, or snap clocks. It does not score reading speed or line length. It does not remove hallucinations from an automatic caption file; you delete those cues yourself. Whisper-shaped junk has its own checklist on Whisper SRT to Word.

A file that is empty, or that has no cues, stays in the preview as a message rather than a fake transcript. A file past the size limit is refused, and the converter states the limit. UTF-8 is what the upload reads. A leading byte-order mark is ignored. A UTF-16 file or a legacy code page will look corrupted; re-save as UTF-8 and drop it again. This site does not transcode encodings.

Related jobs stay on their own URLs so this one can stay the reading transcript. What an SRT file is walks the cue line by line. Transcript to Word is the map for Zoom, Teams, Otter, YouTube, Whisper, and Descript exports. The guide index lists the rest.