Captions
Captions timed word by word from the transcript, in four styles, burned into the picture; SRT and WebVTT out and in.
Captions are made from a transcript (see Transcripts and text-based editing): the timeline's words are grouped into captions, and each caption is an ordinary text clip with wordTimes (each word's start and end), so its animation follows the speech word by word, and caption: true, so captions can be found, replaced and exported as a group. Because they are text clips they render the same way in preview and export (burned into the picture), and the Inspector restyles them like any title.
| style | look |
|---|---|
subtitle (default) | plain text on a dark bar, low on the frame, up to two lines |
karaoke | bold outlined text; each word lights up as it is said |
highlight | big uppercase outlined words, the one being said in a colour (highlightColor) - the short-video look; up to five words |
pop | one to three big words at a time, each popping in as it is said |
A new caption starts at a pause of 0.6 s or more, after a sentence, after a comma once it has three words, or when it would run too long (lines, words or seconds). Lines are as long as fit the canvas at the style's size: sizes follow the canvas's shorter side, so a 9:16 reel gets narrower captions than a 16:9 video. Each caption stays about a quarter of a second after its last word, never into the next one, and starts and ends on whole frames.
They go on the video lane captions are already on; otherwise on the top video lane when nothing is there while they run, else on a new lane above it. Making captions again replaces the ones there. Make them after the cutting is done: a cut through a caption shortens it but doesn't re-time its words, so make them again after you change the edit. The editor says so for a moment whenever an edit shortens a caption.
In the editor, the Captions row of the Transcript tab: pick a style, then Add captions (one undo step). SRT and VTT download the timeline's captions as a subtitle file; Import… places the captions of an SRT or WebVTT file instead, re-broken to the style's line length, with their words spread over each caption. Select a caption to restyle it in the Inspector: the Highlight colour sits under Animation, and a note says its words follow the speech.
Agents use fablecut_captions:
{}or{style:"highlight"}makes captions from every clip with sound (transcribing what has no transcript yet, asfablecut_transcribedoes), and answers with where they went and the first few.clipIds,from/tolimit it;lines,wordsandsize(a scale on the style's size) adjust it;trackpicks the lane;replace:falsekeeps the captions already there.{export:"srt"}or{export:"vtt"}writes the captions that show (not disabled, not on a disabled track) toexports/<project name>.srtor.vtt, adding a number when that name is taken (<project name> 2.srt), and returns the path: use that one.{import:"<path to .srt or .vtt>"}places a file's captions (the file's text works too).{remove:true}deletes every caption clip. Locked captions are left alone and refuse the call unlessforce:true.- Check one with
fablecut_frame {time}while a word is being said. Restyle them all withupdateClip(e.g.props:{highlightColor:"#22d3ee", y:500}) on each caption id from the compact view, which lists a caption as[5 words timed to speech]instead of its times.