On Tue, Mar 10, 2009 at 11:10 AM, Marcus D. Leech <address@hidden>
OK, so I decided to use the averaging method, rather than the maximum
method. It produces reasonably good looking plots:
True, not bad. One surprise, though -- what's that notch around 1420.5?
It definitely has an "averaged" look to it -- much like what you'd expect from time smoothing and not frequency necessarily. One thing that low-level heuristic grass might give you is a feeling that the relatively flat segments are alive at least, sort of like comfort noise during vocoder silence. If that matters.