What this tool does
Only the dialogue in a subtitle file matters to a reader; the rest is structure meant for a player — cue numbers, timelines, style tags. This page removes the structure and keeps the words.
- Read it as an article — output continuous paragraphs, broken at pauses and sentence-ending punctuation.
- Use it as a list — output one line per cue, one subtitle line each, handy for a dialogue sheet or for feeding another tool.
- Clean as it goes — SDH cues, watermarks and duplicated neighbouring lines are removed by default.
- SRT / VTT / ASS / SBV / LRC / TXT — all six formats are recognised.
Drop it in, not just upload
The "1 Add subtitles" column on the left accepts an uploaded file, and you can also paste subtitle content straight into the text box. If you picked the wrong file, click "Clear" and start again.
This page reads files as UTF-8. If the subtitles open as a wall of question marks or boxes, the encoding is broken — convert it to UTF-8 in Garbled text repair first, then come back.
Choosing between the two layouts
At the top of "2 Extraction options" is the layout: Continuous paragraphs breaks text at pauses and sentence-ending punctuation so it reads like an article; One line per cue gives one line per subtitle entry, matching the entries of the original file one to one.
Copying dialogue for notes, practising translation or reading it as prose calls for continuous paragraphs; building a dialogue sheet, comparing line by line with the original or feeding another tool calls for one line per cue.
What each of the five clean-ups removes
Remove SDH cues: deletes bracketed sound descriptions such as "(footsteps)" or "[music]" and keeps the dialogue.
Remove watermarks and ads: drops lines such as fansub credits, URLs and channel handles.
Remove duplicate neighbours: when the same sentence appears twice in a row (a playback glitch) only one copy is kept; duplicates further apart carry meaning and are left alone.
Remove speaker prefixes: "JOHN: Hello" becomes "Hello". Off by default — script-style text needs to show who is speaking.
Skip lyric lines: lines that are entirely "♪ … ♪" are skipped. Off by default — lyrics are words too.
Taking the result away
The "3 Result" column on the right shows the extracted text live and recalculates as soon as you change an option.
When you are happy, click "Download" to save a .txt or "Copy" to take the text with you. For languages that do not separate words with spaces, such as Chinese and Japanese, line merging follows different rules, and the tool handles that per language automatically.
What is left untouched
The dialogue itself is not rewritten. The clean-up switches only decide which whole lines or blocks are dropped; they never change the wording of an entry.
Timelines and cue numbers are simply not output, not altered and then output — the original file is untouched, and this page does not offer an export with timelines. That job belongs to Format conversion.
Common questions
Do ASS style tags stay in the text? No. ASS `{}` override blocks are stripped and only the dialogue remains.
Does a bilingual file come out as two lines or one? As many lines as the entry had; this page neither merges nor splits them — to separate them use Bilingual split.
Why are speaker prefixes kept by default? The name in "JOHN: Hello" is often useful information (scripts, dialogue sheets), and removing it makes it hard to tell who is talking, so the choice is left to you.
Can LRC lyric files be extracted too? Yes. LRC time tags are removed and the lyric lines remain; if you do not want lyrics, tick "Skip lyric lines".
Privacy
Subtitles are processed entirely in the browser: nothing is uploaded and nothing goes through a server. The text stays on the current page only and is cleared on refresh.
Related tools and guides
To understand what a subtitle file actually contains see What is inside an SRT file; to translate the extracted dialogue go to Subtitle translation; to turn it into a table use Subtitles to Excel/CSV; and run Subtitle QC before handing anything over.