I was commiserating with my friend and colleague R. on the phone this week about the difficulty of laying intelligeable voice over a huge, thick, turgid bed of audio oatmeal, courtesy of an unnamed artist who provides some incidental music. R. was interested in looking into spectral ducking as a way to solve his problem.
My first suggestion was that he use Soundhack for the purpose. In case you're not familiar with it, fire up a copy of Soundhack and try this:
- Open an audio file
- Select "Spectral Dynamics" from the Hack menu
- Check the "Use soundfile for thresholds" box at bottom
- Click on the button to choose a source file.
Presto. But how about doing something using MSP? After a little cruise through the MSP tutorials, and some fun with the tutorial and help example patches, it was starting to take shape. In addition to problems being the irritants which can be used to produce pearls, they're also an opportunity to chat with friends. And who better to touch base with but Richard Frigyes Fülöp Tivadar Dudas, Cycling '74 gourmand, bon vivant, and--as his initials might imply--precisely the sort of person you'd wish to be in touch with. Some discussion, some rumination, and some time yields the following Max patch and accompanying pfft~ subpatcher which (alone or in combination with further tweaking) should make for some fun. Just cut and paste the following text into two MSP patches:
Here's the top-level patch, ducking_it
And here's the pfft~ subpatch, spectralduck_pfft:
Of course, there's plenty of fun to be had beyond this simple outline (thanks again for your help, Richard!). For example, I've spent some pleasant time using a variant of this patch to process a file and to duck it using the same source material, delayed by a selectable amount. See what you can do with it.