Turn subtitle files into plain text
Extract text and remove timestamps and cue numbers from one UTF-8 SRT or VTT file up to 2 MiB. Keep blank lines between cues or join them into one paragraph, then copy or download the TXT result. Optional deduplication removes only identical adjacent cues, so partial repeats in rolling captions stay.
Source subtitles
One UTF-8 file · up to 2 MiB · 20,000 cues
The example is synthetic. The tool does not save a history. Download your result before leaving or refreshing.
Choose the text layout
Removes timestamps and supported formatting tags. Existing words stay in order; no summary is generated.
Your output will appear here after a successful preview. Changing the source or options clears the previous result.
Worked subtitle example
Both outputs keep all three subtitle lines in order. Keeping cue breaks inserts blank lines. Joining the cues makes a single paragraph.
1 00:00:02,000 --> 00:00:04,500 Keep the original subtitle file. 2 00:00:05,000 --> 00:00:07,250 Check timing in your final player. 3 00:00:08,000 --> 00:00:10,000 Download the reviewed copy.
Keep the original subtitle file. Check timing in your final player. Download the reviewed copy.
Keep the original subtitle file. Check timing in your final player. Download the reviewed copy.
Download the equivalent WebVTT input · Example and validation notes
From subtitle file to a reviewed copy
Start with a subtitle file
Choose SRT or VTT, or paste the original timed text. To extract words from a recording instead, use the transcription tool.
Choose the reading layout
Keep cue breaks when checking against the subtitles. Join the text into one paragraph when moving the words into a document. Supported formatting tags are removed; WebVTT voice labels become text prefixes.
Review the result
Keep repeated lines by default. Optional deduplication removes only identical adjacent cues; it does not merge partial rolling captions, restore sentences or summarize the content.
What this tool reads
One UTF-8 SRT or WebVTT text file up to 2 MiB and 20,000 cues, with timestamps below 100 hours. Empty cues, invalid times and out-of-order starts need correcting before export. Overlaps are kept with a warning. Streaming timestamp maps are not supported.
Processing runs in this browser without uploading subtitles or downloading a speech model. The workspace has no saved history. The site still loads application resources; optional analytics and submitted feedback follow the Privacy Policy.
Common questions
Can I turn subtitles into a transcript without the recording?
Yes, if you already have an SRT or VTT file. This tool extracts its text; it cannot recover words missing from the subtitles or verify them against the speaker.
Will it remove duplicate sentences?
Only if you enable the option, and only when adjacent cues have exactly the same extracted text. Partial repetitions and non-adjacent lines remain. Repetition may be intentional, so review it before deleting.
Does it create a summary or identify speakers?
No. It keeps the existing words in order. Existing WebVTT voice labels become text prefixes, but it does not identify unlabelled speakers or create a summary.
Continue your subtitle workflow
Turn caption fragments into readable text
Format references: W3C WebVTT draft and Library of Congress: SubRip. This tool handles the subset described above; it is not a full subtitle editor.
