I'm pretty sure that's what Schwa's "humanizer" filter does.
What would be extra useful would be to have three separate things:
1: Constant random-modulation.
2: sample-hold modulation
3: Most important- Random pitch based on the same noise, but with the pitch only varying for each new note. That is, while a single keys is depressed, the pitch does NOT change (or not much.) The notes are only (slightly) out of tune in relation to each other.
This tends to me more realistic for many instruments.
|