Skip to content

Latest commit

 

History

History

arst

    ARST                    12 May 2014

This is a small program for annotating a video file, in particular
labeling individual spoken words in spans of its audio. It's for weird
special projects, but maybe could be used for something normal, like
generating training data for speech recognition. Assumes 64 bits and
lots of RAM.

Best documentation for the key commands is in the source, but:

Space toggles play mode
Up and Down jump to the previous and next interval.
, and . decrement and increment the frame.
/ splits the interval at the current point.
s saves the script.
enter modifies the text in the current interval. If the text contains
spaces, then it splits the interval equally and assigns a word to each.
(Typically I use this by making an interval for a spoken phrase,
fine tuning it in loop mode, and then replacing it with the phrase 
that's spoken, then fine-tuning the breaks between words. Pretty fast
once you get some practice listening for the ends and beginnings of
phonemes.)

Tab enters loop mode. When in loop mode, the current interval is
looped indefinitely. The keys h,j adjust the left end of the interval
(it can't go any further than the next interval, though) and k,l
adjust the right end.