[Audacity-nyquist] Help with Commentator Plug-in

[Matthew W. <mat...@gm...>]

I need help with a project I started.  I am attempted to create a utility to
adjust audio levels on a commentary track.



I am creating this project to use with rifftrax.com.  Here you can download
mp3 commentaries by former Mystery Science Theater 3000 host Mike Nelson and
sync them up with DVDs.  Thus he make fun of movies he doesn't have the
money to license.  Alot of people are authoring DVDs with Rifftrax, which
requires combining the original DVD audio with the Rifftrax mp3.  Most
people use audacity for this.  The problem is that setting volume levels is
extremely difficult.  Most movies change volume considerably between
dialogue and action scenes.  Getting the commentary to be hearable during
loud scenes are not blare in other scenes is a chore.  I noticed that most
real commentary tracks reduce volume when the people talk.  The rifftrax
commentary is usually a few seconds to a minute of silence and then a quick
riff.  Reducing the volume during all of these in a movie is ridiculously
time-consuming.  SO, the project is to create a plug-in or utility that will
reduce the volume on one track during non-silent timeframes of another
track.



I've done some research into Nyquist and it seems that my basic process will
work.  Perform silence analysis on one track and record timeframes, create
envelopes during non-silent timeframes, apply envelopes.  However, I'm still
trying to figure out the limitations of the Audacity plug-in system.



Is it possible to pass multiple tracks to the plug-in system and
differentiate between them?  If so how.  And can I return changes to only
one of the tracks.  I thought it might be possible to have control values to
specify which track is the commentary (0 or 1, the first or second selected
track), whether either one is stereo (0 for mono, 1 for stereo), and process
the s list from there.  Is this feasible? Is there a better way? How is the
order of tracks organized when you pass selections to a plug-in?  Is it the
order selected or the order on screen?



Another thing is tips on how to make this process memory efficient.  These
are going to be VERY large tracks to be processed.  Ineffecient memory use
could easily crash most systems.  I'm not familiar enough with Nyquist to
know how it handles memory.  I don't know which actions create deep copies,
which do reference copies, etc.



Another thing, if this project would not be feasible with Audacity, then I
will probably have to implement it with Nyquist.  If that is the case, how
could I set up a graphical interface to a nyquist script and make it easily
distributable.



I am an experienced programmer, and I have a little background in audio
editing.  Any help you people could give would be most appreciated.