Remove Non-Speech Information from Subtitles
Or paste subtitle text
Options
One box, five choices, and each choice has a page of its own. Brackets removes paired [square] asides, parentheses removes paired (round) ones, speaker removes an all-capitals name and the colon after it at the start of a line, and music removes the note symbols. All four at once is the choice called All. Every one of them edits the words themselves rather than the styling, so the line gets shorter — and unlike the styling pages, an annotation that was removed cannot be put back by the tool that reads the result. This page fills the box in with the choice its title names and lists no other.
Which frame rate the timings are counted in. On the format pages this is what MicroDVD needs, because it stores timing as frame numbers. On the frame rate pages it is the rate the file was timed against — the version the subtitles were made for — and the page's second box takes the rate of the video you are going to use them with. 23.976, 24, 25, 29.97, 30, 50, 59.94 and 60 are the common ones. Getting it wrong shifts the whole timeline, and on the frame rate pages it shifts it by a ratio rather than by a fixed amount.
This page deletes everything inside a paired set of round brackets. `(whispering) I can hear you.` comes out as `I can hear you.`, and `She left (quietly).` as `She left.` The cues, their timecodes and their numbers are the file's own and are not touched.
Round brackets are how a hearing-impaired track marks the things a viewer can hear but nobody says. In British and European practice these are collected under auditory cues: the sigh, the door, the phone, the whisper, the music under the scene. In American practice the same content is filed as non-speech information, the umbrella term for everything on screen that is not dialogue. The two names describe the same bracket, which is why this page answers to both and does not treat one as a translation of the other.
The criterion here is shape, not meaning, and that is worth knowing before you run it on someone else's file. `(sighs)` is a remark about delivery; `(maybe 6)` is a hedge about a price; `(1984)` is a year; `(V.O.)` is a voice-over note that some houses put in round brackets and some in square ones. All four are paired round brackets and all four are removed. Telling them apart needs a reader who understands the sentence, and this page does not pretend to be one — a version that guessed would be quietly wrong on exactly the lines where being wrong matters.
So the page is for files where the brackets are house style rather than accident: the ones your own captioning pass wrote, or a vendor's whose conventions you know. If the brackets came from somewhere else, run a sample through and read the output before you run the season. Nothing is uploaded, all twelve formats are accepted, and a file whose remarks are in square brackets comes back byte for byte — that shape belongs to the bracketed text page.
Examples
A delivery remark before the line and after it
Input
1
00:00:01,000 --> 00:00:03,000
(whispering) I can hear you.
2
00:00:04,000 --> 00:00:06,000
She left (quietly).
Output
1
00:00:01,000 --> 00:00:03,000
I can hear you.
2
00:00:04,000 --> 00:00:06,000
She left.
The two places a remark turns up in a real file: opening the line, and tucked in before the full stop. Both go, and the spacing closes up in both directions. The second cue is why a space in front of a full stop is cleaned too — without that the line would read `She left.` only by luck, and a longer sentence would come out with the space still in it.
A paired bracket that is not a sound at all
Input
1
00:00:01,000 --> 00:00:03,000
It costs 5 (maybe 6) dollars
Output
1
00:00:01,000 --> 00:00:03,000
It costs 5 dollars
This is the case to look at before using the page on somebody else's work. The brackets are perfectly paired, so the page cannot tell that the words inside are an aside about a price rather than a sound in the room, and the hedge is deleted along with them. Nothing in the output marks where it went.
An unclosed bracket and a file in square brackets
Input
1
00:00:01,000 --> 00:00:03,000
(whispering I can hear you
2
00:00:04,000 --> 00:00:06,000
[door slams] I heard that.
Output
1
00:00:01,000 --> 00:00:03,000
(whispering I can hear you
2
00:00:04,000 --> 00:00:06,000
[door slams] I heard that.
The file comes back byte for byte, and each line has its own reason. The first bracket is never closed, so where the remark ends is unknowable and the line is left as it is. The second line's remark is in square brackets, which this page does not pair at all. Both are the page declining to act rather than acting on a guess.
Frequently asked questions
What counts as non-speech information on this page?
A paired set of round brackets and whatever is inside them. That is the whole rule. It covers the sounds a hearing-impaired track describes, such as a sigh, a door or a phone, and it equally covers an aside that has nothing to do with sound, because the page matches a shape rather than reading the sentence. If you need the meaning checked, that is an editorial judgement and no text tool can make it for you.
What is the difference between auditory cues and non-speech information?
They are two names for the same thing, used on different sides of the Atlantic. British and European subtitling practice talks about auditory cues: the sounds a deaf or hard-of-hearing viewer cannot hear and therefore needs written down. American practice files the same content under non-speech information, which also takes in anything else on screen that is not dialogue. Both describe what sits inside the round brackets this page removes.
How do I stop it removing an aside I want to keep?
By not running this page on that file, or by taking the aside out of brackets first. The page has no way to tell a remark about delivery from a parenthesis about a price, so both are removed. If your file mixes the two, the honest options are to fix the brackets at the source or to use the bracketed text page instead, which pairs square brackets and will leave your round ones alone.
Does the removal change the cue timing or the cue count?
The timing is untouched: every cue keeps its start and end, and the numbering stays as the file wrote it. A cue disappears only when every line of it was a bracketed remark, since there would be nothing left to display, and the report counts those separately from the cues that were merely edited. Styling ranges outside the removed text move with it and survive; a range that covered the remark itself has nothing left to point at and is dropped.
Why is there a space left before a full stop in my output?
There should not be. When a remark sits in the middle of a sentence, deleting it leaves the space that came before it stranded in front of the punctuation, so the page removes that space as well. The marks it will close up are the full stop, the comma, the ellipsis and closing brackets; it deliberately leaves the space in front of a question mark or an exclamation mark alone, because in French subtitles those are correct with a space in front of them.
Which subtitle formats does this work on?
All twelve the site reads: SRT, WebVTT, ASS, SSA, SBV, SAMI, LRC, MicroDVD, TTML, JSON, plain text and CSV. The remark is in the cue text and every one of those formats has cue text, so none is excluded. The format is worked out from the contents rather than the extension, and you can override it in the queue if the file is ambiguous.