[Audacity-devel] New transcription toolbar functions

[Shane M. <smu...@ob...>]

For the curious, I have added two new functions to the transcription
toolbar: "select sound" and "select silence".  These extend the
selection to the region of silence/sound surrounding the current
selection.  Its intended use is: click on a sound; hit the button; the
entire 'word' is now selected.  I think this would be really cool if it
were bound directly to double-clicking, very analogous to how an entire
word gets selected if you double-click on it in a word processor; but
that would probably require a smarter sound/silence detector than I
currently have, like one that doesn't need to be calibrated to noise.

I would actually like to make the sound/silence detector selectable, and
have numerous different versions.  I know people have used cepstrum for
this, and I imagine that a really cool one would use the pitch
detection: the change is something like the cosine of the angle between
the autocorrelation vectors at any two points.  This could be a good way
of picking out sounds in continuous audio situations. Some type of
fourier-domain analysis akin to the noise remover would also be nice to
have. If anyone has interests/skills in this area (and there are a bunch
of free software libraries that do things like this already), it would
be great to try out some different things.

Stm...