Processing drums in a studio (quantisation, etc) - How should I go about it?

Reggae_Mangle

Silver Member
How would I process this take in a studio?

https://soundcloud.com/jimmy-rage/how-should-i-process-these-drums-quantisation-velocity-etc

I'm specifically looking for advice in terms of things like quantisation or any other processing I should be doing in terms of timing, velocities and the feel of the track. I'm not too worried about the sound since it was recorded using an e-kit and I can change the kit completely in Superior Drummer 3.

I took the advice of some of the guys here and have been recording multiple takes in my studio for songs that I aim to put on a demo album.

But I am no Derek Roddy, and while I haven't tried playing guitar over the track to see how everything lines ups, there are definitely some sore spots I noted.

I would like to keep things sounding as organic as this though: no hard quantisation or excessive editing. I am particularly open to the idea of recording additional takes and comping things together.

It should be noted that I played to a click, but I played ahead of the click, which was at 210bpm. Are there any problems I should anticipate on account of not being locked into the grid like a pro would be? I think I could definitely lay down guitars over this take, but I wouldn't want to gloss over the errors... It's important that I fix them before I get down to tracking guitars.

Any help would be appreciated.
 
Last edited:
Couple notes...

If you're laying down a track with the intention of quantizing it, it needs to be as close to the grid as possible.

If you have multiple tracks, it's often beneficial to quantize them all at once instead of each individually.

If a quant task is daunting (a restoration from a multitrack analog source), break it up into smaller tasks.. verse, chorus, bridge for example.

If you're new to this, start simple. Record 8 bars of money-beat followed by 8-bars of similar chorus. Quantize it.
 
It's a good question given the sound source being electronic. I thought the extended 16th section was good, although the kick should come down about 6 dB in the double bass sections IMO. You can do that via velocity, automating the volume of the track, or using a compressor to raise the volume of the non-16th sections. Less velocity will also make it sound more natural and less needing of quantization, the sound right now sounds like max velocity and accentuates every inconsistency in the double bass. The snare also should come up in volume, it could use a higher velocity if possible because you really want a rimshot sound for this.

Honestly other than maybe the first part with the 8th note double bass - it sounded a bit late consistently, you could try just moving the bass drum track forward or punching in that part - I think most of this is totally fine for a demo album with some further mixing.
 
If your drums are consistently and reliably ahead of the click, you might want to nudge them backward a bit to make them closer. The closer they are to the click to begin with, the easier and more accurate the quantizing process will be. That said, if they're sometimes ahead and sometime on-center, and sometimes behind, then leave them alone and let the software do its thing. You'll most likely end up manually correcting some hits anyway.

Quantize the whole track, and make note of where the software "messes up" the tracks. Undo the quanitization process, then go in and manually fix those spots, until you can process the whole length of all the tracks at once. When the whole length of all the tracks can be processed at once, undo the quantizing, and copy these tracks to another playlist/session/whatever. Then, quantize the entire length of the tracks. If you accidentally quantize too much, you can go back to the unquantized (but manually corrected) tracks and start over.

As for how accurate, your ears will tell you. Do a small section, lay down a scratch guitar track, take a break, and then come back and listen.
 
If your drums are consistently and reliably ahead of the click, you might want to nudge them backward a bit to make them closer. The closer they are to the click to begin with, the easier and more accurate the quantizing process will be. That said, if they're sometimes ahead and sometime on-center, and sometimes behind, then leave them alone and let the software do its thing. You'll most likely end up manually correcting some hits anyway.

Quantize the whole track, and make note of where the software "messes up" the tracks. Undo the quanitization process, then go in and manually fix those spots, until you can process the whole length of all the tracks at once. When the whole length of all the tracks can be processed at once, undo the quantizing, and copy these tracks to another playlist/session/whatever. Then, quantize the entire length of the tracks. If you accidentally quantize too much, you can go back to the unquantized (but manually corrected) tracks and start over.

As for how accurate, your ears will tell you. Do a small section, lay down a scratch guitar track, take a break, and then come back and listen.

That trick with quantising to identify problem spots is a great idea! Thanks! I did some vocal guitar to check that everything was in order, there are definitely some spots where a hit came too early or too late, or the double bass is slamming. Those are the areas I really want to fix, so I will try this method, thanks!

It's a good question given the sound source being electronic. I thought the extended 16th section was good, although the kick should come down about 6 dB in the double bass sections IMO. You can do that via velocity, automating the volume of the track, or using a compressor to raise the volume of the non-16th sections. Less velocity will also make it sound more natural and less needing of quantization, the sound right now sounds like max velocity and accentuates every inconsistency in the double bass. The snare also should come up in volume, it could use a higher velocity if possible because you really want a rimshot sound for this.

Honestly other than maybe the first part with the 8th note double bass - it sounded a bit late consistently, you could try just moving the bass drum track forward or punching in that part - I think most of this is totally fine for a demo album with some further mixing.

Thanks, that's really helpful. Yes, a lot of the notes are max velocity, I'll see if I can set a ceiling to improve things.

Couple notes...

If you're laying down a track with the intention of quantizing it, it needs to be as close to the grid as possible.

If you have multiple tracks, it's often beneficial to quantize them all at once instead of each individually.

If a quant task is daunting (a restoration from a multitrack analog source), break it up into smaller tasks.. verse, chorus, bridge for example.

If you're new to this, start simple. Record 8 bars of money-beat followed by 8-bars of similar chorus. Quantize it.

If I could do without quantisation, I would, but alas, my playing is not at that level yet. I need to nudge stuff a bit closer to the grid. At the same time, I don't want to make it sound mechanical, the less quantisation, the better for me. I don't mind a little push-pull in my music, but I don't want wild gyrations in tempo either.


How should I go about quantising? A weak quantise? Or should I permit some swing in the notes instead to achieve that end?
 
How should I go about quantising? A weak quantise? Or should I permit some swing in the notes instead to achieve that end?

In my world, there are only two modes of quant...

1: Fixing the single mistakes (A single BD hit that's off or a fill that drags)
2: Select-All, Select Quantize, Select Auto

There is no "option 3" where I sit down to fix the entire song by hand. Doing that is about as exciting as knitting a sweater. If I can't play the part, I simply practice more and/or slow down the tempo a few BPM till I can hit it. After I record, I bump the tempo back up on the recorded part. No sense wasting a bunch of time if I'm within several BPM of the goal.
 
I experimented a bit with some quantisation this afternoon. I found that I get extremely professional results by setting the quantisation strength to around 20% with zero swing in logic. I only quantised the kick drum and was extremely pleased with the results. I didn't save anything though, as suggested by brentcn. I'll keep the original tracks and process later at one go once I have everything laid down, including a trim of maximum velocity to 120. Thanks, everyone.
 
There are two tools that make me sound like a pro: Auto-Align and Studio One’s time stretching algorithm.

  1. Set up Studio One to loop over the tune and lay down multiple takes.
  2. Group the tracks, which allows me to manipulate all the tracks as a whole.
  3. Select the best take and make notes on which fills or mistakes need to be replaced by another take.
  4. Apply Auto-Align using the snare as the source for all transients (this is debatable, some use an overhead).
  5. Set all the tracks’ time stretching algorithm to DRUM.
  6. Set the quantize guide to the kick and snare.
  7. Analyze all the tracks and evaluate the results
  8. Click APPLY
  9. Adjust a few nodes (it’s never perfect)
  10. Sit back and marvel at how vastly better it sounds with drums actually playing on the beat.
  11. Sip cold beer.

The tedious part is replacing poorly played sections. The time stretching is not applied to any part I will use to replace a section with, and must be done to each of those sections. Auto-Align is not done because it’s universally applied to each track.

Auto-Align

When I first tried time stretching in Studio One I was expecting nasty sounds, but that’s not the case. It has three different algorithms for time stretching, one for drums, two for harmonic instruments, and it’s almost too easy to use.

That said, I’ve not used or tried Pro Tools (the “Photoshop” of DAWs).
 
There are two tools that make me sound like a pro: Auto-Align and Studio One’s time stretching algorithm.

  1. Set up Studio One to loop over the tune and lay down multiple takes.
  2. Group the tracks, which allows me to manipulate all the tracks as a whole.
  3. Select the best take and make notes on which fills or mistakes need to be replaced by another take.
  4. Apply Auto-Align using the snare as the source for all transients (this is debatable, some use an overhead).
  5. Set all the tracks’ time stretching algorithm to DRUM.
  6. Set the quantize guide to the kick and snare.
  7. Analyze all the tracks and evaluate the results
  8. Click APPLY
  9. Adjust a few nodes (it’s never perfect)
  10. Sit back and marvel at how vastly better it sounds with drums actually playing on the beat.
  11. Sip cold beer.

The tedious part is replacing poorly played sections. The time stretching is not applied to any part I will use to replace a section with, and must be done to each of those sections. Auto-Align is not done because it’s universally applied to each track.

Auto-Align

When I first tried time stretching in Studio One I was expecting nasty sounds, but that’s not the case. It has three different algorithms for time stretching, one for drums, two for harmonic instruments, and it’s almost too easy to use.

That said, I’ve not used or tried Pro Tools (the “Photoshop” of DAWs).

That sounds pretty cool. I don't have Studio One though, just Logic and Cubase.

I read the web page and it didn't seem like you could do all you do though! Interesting that it has algorithms too...

How common is it to treat drums like that though? I've heard horror stories about some drummers who can't keep time... But what about the pros?
 
Last edited:
That sounds pretty cool. I don't have Studio One though, just Logic and Cubase.

I read the web page and it didn't seem like you could do all you do though! Interesting that it has algorithms too...

How common is it to treat drums like that though? I've heard horror stories about some drummers who can't keep time... But what about the pros?

Logic has a quantize tool. I don't know about Cubase.

After taking a few workshops at Sweetwater, the consensus is that if the band and/or producer wants the drums locked to the grid, then it's done. No matter who the drummer is.

For high-tempo metal drumming, I believe all the instruments are locked to the grid, otherwise it'd sound pretty sloppy.

However, if you watch the documentary of Tom Petty & The Hearbreakers (Netflix), you'll see that when they got signed and went into the studio, the producer rejected Stan Lynch and opted to use a studio pro. Stan was bummed/dejected (as anyone would be), but Petty didn't like the feel of the other drummer's playing. Stan Lynch was brought back and the rest is history. His drumming has a great feel, and none of those early records were recorded to a click track.
 
Logic has a quantize tool. I don't know about Cubase.

After taking a few workshops at Sweetwater, the consensus is that if the band and/or producer wants the drums locked to the grid, then it's done. No matter who the drummer is.

For high-tempo metal drumming, I believe all the instruments are locked to the grid, otherwise it'd sound pretty sloppy.

However, if you watch the documentary of Tom Petty & The Hearbreakers (Netflix), you'll see that when they got signed and went into the studio, the producer rejected Stan Lynch and opted to use a studio pro. Stan was bummed/dejected (as anyone would be), but Petty didn't like the feel of the other drummer's playing. Stan Lynch was brought back and the rest is history. His drumming has a great feel, and none of those early records were recorded to a click track.

That's an interesting story. It is a different genre though, so yes, I think I will have to figure out how to quantise without ruining the feel. Thanks for the input.
 
Back
Top