Refining the drum track
Let's say I have a project, built by a template that follows my work flow, a tempo for my project, a timeline with markers indicating the various points and movements of my song, a scratch guitar track, and scratch vocals.
Often, it's easier for me to get a good run through on guitars if the click track I am playing to isn't actually a click track at all. Instead of a lifeless robotic click noise denoting the downbeat, I'll replace it with a Battery drum track that drops a kick drum on the down beat. It gives my playing a nice little backbeat to play along with.
Sometimes I already have a good indication in my head as to the basic beat of the song. At this early stage it's about getting the general idea down enough to allow me to continue forward, but not getting bogged down in the details. And I almost always find programming drums to be easier done with live performance as opposed to straight up MIDI note editing.
...which is kind of odd when I consider the years I spent programming house music under the Raygun moniker. Almost every note in that project is created manually editing MIDI piano rolls as opposed to live rhythm performance. It's simply a different approach lending itself to a different sound. House music is by nature electronic and very robotic, so having a human feel can actually be frowned upon. Not in every case, but non-quantized beats can be tough for a DJ to manage during a live set, so it's definitely a consideration that is followed rather closely.
The music I produce as Yellowgold is always far more organic... at least it is in my head. The problem is always the simple fact that A, I am not a drummer.... and B, I don't have a drum set to try to become one. I'd certainly love to be able to do it, but from a space and noise perspective, it's simply not that possible.
So my approach with MIDI drum programming usually follows like so:
1. Click track or very basic 2-4 bar loop of programmed drums. As simple as a kick snare combo with some super light hi hats. I don't want the drums to be too detailed at this early stage, I merely want a backbeat that I can groove to while I play. If I create something too creative, I might end up accenting certain parts of the beat that might clash with the other elements I have in mind that hit the track at a later stage.
2. Once I have this, lay down my scratch tracks that give the track a beginning-to-end structure, the skeleton for my entire song.
3. Once I have a sense of the twists and turns, some fills and a more detailed drum sound start to take shape in my mind. Certain parts seem ripe for a particular tom fill. Suddenly crashes on particular downbeats feel appropriate. Once I have an idea for all of these things, I'll create an entirely new region for Battery and start to perform the foundation, live, with the Keystation Pro-88. I've gotten pretty used to the key mapping for drums on the keyboard, so its become much easier for me to play the kit freestyle.
The thing to realize here is I don't expect this pass to be 100% perfect through to the end of the song. I am hoping to nail, with little need for quantization if possible, the feel of the song for around 16 bars, 32 would be great. I want a long enough block of solid drums that I can feel comfortable copying that throughout the rest of the track as need be, once I've done what's next.
4. When I have that solid block, I treat the notes, sometimes on an individual basis, with select quantization. I don't blanket quantize unless I'm lazy. (Hey, sometimes it happens.) And quite honestly, the material doesn't always NEED that kind of attention. But I find that it feels alive if you treat it with more precise attention than it does if you throw a straight up 8th or 16th note quantize onto the whole block of drums. I almost always quantize the downbeat kicks strictly, so the downbeat is always dead on. I usually quantize the hi-hats with a light randomization (somewhere in the 6-8% range) so they don't sound too robotic. Snares usually hit very close to down beat, though I might randomize those around 2-4%. I also take a lot of time refining the velocity of these hits... particularly in the hi-hat line. When hi-hats hit as frequently as they can during a drum session, having them all hitting at the same or nearly the same velocity is a surefire way to spot MIDI drums, so I really try to take some time getting those sounding as natural as possible.
I'm exhausted typing this cause I realize how much time I spend on these things and yes, it's tiring. It's by no means my favorite part of the process, but I feel its essential to producing the kind of sound that I'm looking for. It would be way easier to just do blanket quantization and be done with it, but I'd hear the robotic nature of those drums every time I heard that track and it would eat me up inside.
Up next: What do you do when you've recorded an acoustic guitar as your very first track in a project, to no click track, then later decided you want to make that un-timed track the foundation of a production complete with drums and everything else? I did that for Heavy Bones, and I'll share with you what I did.