I'm wanting a Tensor Flow a.i. trained plugin that automatically massages the timing and dynamics (of midi date) to match the habits of certain famous drummers.
Hats and rides are the hardest to deal with. I don't think round robin works properly in most drum VSTis, in that there is usually a range in velocity that is supposed to represent "drummer variance" but within that range the timbral changes are too subtle versus the volume. In other words, dynamics change, but not the timbre within that range. A closed high hat that gets louder and softer - but sounds basically the same while doing so - doesn't sound real, hats still moving from a previous hit sounds different than just starting from a stand still.
And swing functions need to incorporate a velocity scaling aspect for the same reason. A 16th feel with just timing offsets is not how a real drummer plays. It's bizarre that some VSTis leave out swing entirely.
I suppose swing has been extricated from modern perception, based on observations of new students trying to play anything that is swung or shuffled. Or some drummers that are technically very adept, but are oblivious to the subtle in dynamics of ghost strokes on James Brown funk beats or a Purdie shuffle, or Stewart Copeland time keeping.
I have wasted too much of my life doing tweaks like in Craig's article, there really should be a better solution in the year 2020. Jamstix gets close, but is too monolithic. I want "swing style","swing percentage", "swing dynamic curve", "swing variance beginning/middle/end of measure".. and more.
Last edited by Chip McDonald; 11/06/20 04:07 PM.