Suggestion: a time-based instead of word-based caption buffer

[casperdewith]

Currently, when setting up live captioning (Audiovisual > Captioning), the dialog box asks you to enter a buffer size, which is counted in words.

I would like to be able to choose a different unit: a (real) number of seconds. This means that when this unit is selected, and I chose x seconds, any prefix of my buffer that has not changed for more than x seconds is output.

I request this functionality because some strokes in my steno theory have a lot of words. I obviously want to be able to undo any single stroke without anyone noticing, so I have to use a large buffer. As a consequence, though, the delay between me stroking a stroke and it being displayed is often long. And manually flushing the buffer after every sentence to shorten this delay is not fun.

[greenwyrt]

I can definitely see the usefulness of this feature. The places that have to be changed are the display_captions in the window editor and send_cap in captionWorker. It probably will take me a bit of time to work on this, because the buffer mechanism right now is based solely on the text, and using the time is something different.

I have an idea of an approach and this is more a comment to myself: to start, in display_captions, subset paragraph element collection by last pos to word before cursor, do time diff, use length of elements that pass to move cursor to new position.