Pull the dialogue out of a styled subtitle file and leave the styling and timing behind.
Open a project in the editor and go to Captions. You can start from a video you are captioning, or from nothing at all if you only need the file converted.
Bring in the ASS file. The cues and their timings are read in as they are, so nothing is re-timed and nothing is re-transcribed.
Export as TXT and the timings and cue numbering are dropped, leaving the spoken text as a readable transcript.
An ASS file is the least readable of the caption formats by hand. Each line is a Dialogue event with ten comma-separated fields before the text even starts, and the text itself can be full of override tags in braces controlling colour, position and timing.
Converting to plain text strips the script headers, the style definitions, the event fields and the override tags, and returns the spoken lines. For a typeset file this is a substantial reduction, which is the point.
If the file contains signs and typesetting as well as dialogue, expect those lines in the output too: they are Dialogue events like any other, and nothing in the format marks them as scenery rather than speech.
No. Anything in braces controlling colour, position or timing is markup and is removed, leaving the readable line.
Because ASS stores typesetting as ordinary Dialogue events. Nothing in the file distinguishes a translated sign from a spoken line, so both come through.
Free, in the browser, and the captions can be burned into the video in the same place.