We Can't Go On... (Part 2)
...Beating Like This
...Beating Like This. Chris Meyer and Matt Isaacson with the second part in our series on better drum programming. This month: how to make your percussion sounds stand out from the crowd.
Samplers have been used for replacing drum sounds in recording studios for some time now. In the second part of our series on creative drum programming, we look at how to make your sampler behave more like the percussion instruments it simulates.
WE SPENT QUITE some time in the last issue slamming your typical drum machine as a percussion playback device. Add to this the primitive pattern programming on most drum machines, compared to the current generation of MIDI sequencers, and the fact that more and more musicians are using their "main" sequencer to sequence their drum and percussion parts, and it becomes obvious that we should perhaps look elsewhere for the supplier of our beat.
Let's see - where should we look? Well, what were the problems we outlined with drum machines? Hmmm... Lack of dynamics (volume, pitch, and timbre), no dynamic allocation of voices if we wanted it, length and variety of sounds... Hey! Aren't these things that have come to be expected by keyboard players?
You bet. And if we're truly using another sequencer to play back drum and percussion parts, a sample playback module turns out to be about the best thing there is for our sounds. Playback modules are velocity-sensitive, have a wider variety of sound-modification options (envelopes, filters, and the like) than your typical drum machine, have a wide and easily expandable library of sounds (which can be modified at will), and have a variety of playback modes and voice allocation techniques for layering, velocity fading, velocity switching, and so forth.
If you already have a drum machine, and feel like keeping it around (or it already contains your best work), bless the MIDI jacks that are (we hope) on the back of it - with their help, you can slave sample playback modules off it to double up on or replace its sounds, and eventually transfer your patterns to a sequencer if you so desire.
So, if we want to use a sampler for playing back our drum and percussion sounds, what kind of sampler do we look for? Well, bad news first - the least expensive MIDI'd sampler you're going to find costs what an upmarket drum machine would. Now the good news - the least expensive sampler you can find will sound better and be more expressive than all but the most expensive drum machine. And unless you plan to be playing keyboards as well, or will be using the keyboard to enter your rhythms (a cost-cutting measure for those who are keyboardists and not percussionists at heart), you can do more than just get by with a rack-mounting sampler, which tend to cost slightly less than their keyboard-equipped counterparts. The money you save will be better spent on percussion-style controllers, whether you're a drummer or not.
The criteria to use in choosing a sampler for percussion work break down into four categories: sound (quantity and quality), price (an obvious one), features, and hassle. "Hassle" encompasses how hard it is to use the machine, how big the sound library and support is for the sampler, and the mystical problem of sampler delays. "Features" are what you need to put into practice the various programming techniques that we'll be discussing for the bulk of this instalment - pick out the ones you think you'll be using, then pick out a sampler that has those features on it. Balance them out against "price" when it comes time to purchase. And while you're at it, go ahead and look at some of the brand-new drum machines - they too have caught on to the problems we discussed last month, and are adding features (including sampling) to give us a better beat.
But for now, on to those promised tricks (including how to avoid certain pitfalls). We're going to divide these into two areas - those that affect the way a single sound is heard, and how different sounds or restrikes of the same sound can interact.
VELOCITY REFERS TO how fast something (such as a drum stick, or your hand) moves, which tends to correspond with how hard you strike. One of the amazing little quirks of nature is that the harder you hit something - a drum, your kid brother, whatever - the louder it sounds. It also tends to change in tone and pitch (your brother's voice may get higher and more strained, for example). Our desire is to translate, or as it were, properly interpret, this quirk so as to have more realistic-sounding drum and percussion voices at our disposal.
Let's start with that "harder equals louder" effect. Velocity modulation of the amplifier (also known as the "VCA" level, and virtually the only dynamic control available on drum machines) is standard on all but toy samplers. The sampler version also happens to be more useful. Sampler velocity resolution is greater, with up to 128 discrete levels resolved - most drum boxes offer 16 or fewer levels, and some have only a cumbersome two-level normal/accent option.
Some samplers also allow subtle (or blatant) modulation control by note dynamics. This is important, because it is often preferable to apply different amounts of velocity modulation to different sound parameters. More on this later.
Real percussion instruments tend to vary in timbre, as well as in volume, when they are struck with different velocities, so it's useful to have dynamic control over a sampler's timbre, too. Time, then, to look at several common methods of achieving this.
If your sampler offers dynamic filtering (most do), velocity modulation of the filter cutoff frequency is a good starting point. When the sampler is programmed so that a higher note velocity opens up the filter, louder sounds also sound brighter, with more edge to the attack and more meat revealed in the rest of the sound. This is usually presented as velocity modulation of the filter envelope amount, since the filter cutoff sets the lowest or average filter frequency from which the filter envelope shoots up. Samplers without envelope-able filters, such as the Akai S900 and Roland S50, tend to use velocity to modulate the cutoff directly.
Let us not belittle the filter cutoff frequency - in addition to providing a simple control over timbre, it's helpful in masking shortcomings of the sample playback mechanism. In particular, where it is possible to have VCF cutoff track the sound's pitch, you have a method for covering up the clock noise which becomes audible in most samplers when the pitch of a sound is transposed down far enough.
Another shortcoming of samplers, quantisation noise (that rising hiss as the sound dies away), can be hidden with the normal filter envelope by starting out high for the attack and fading down towards the end of the sound to mask the hiss. This is especially helpful with toms - it allows them to retain a fat roundness with all of the grit of the attack, so long as the controls are set right.
A warning. It often becomes a plate-spinning act to prevent velocity from uncovering all of the nasties you were trying to hide by lowering the filter cutoff and envelope amount. Samplers that route velocity to the filter envelope amount instead of to the filter cutoff are much easier to control in this respect. But be wary of samplers that do not have controllable filters on their output stages: while some will tell you (truthfully) that the absence of such filters cuts down on the noise in the output, we often need them for improving the lot of our problem samples.
In the same way that the filter envelope can be used to mask noise at the end of a sample, a sampler's VCA envelope can be used to tidy up samples which cut off before the original sound was able to die out completely - a useful technique if you find yourself running short of sampling memory. We're only offering it as a cure to an undesirable situation - remember that one of our main complaints about drum machines was the shortness of their sounds. In general, you will find that the longer the sample you can afford to take, the more realistic it will sound.
Of course, envelopes do have purposes in life other than fixing faults - they can be used creatively. Sounds that are lacking in a proper percussive envelope (such as something that was over-compressed, or a sound that was never intended for percussion in the first place, like a steam blast) can now have such an envelope imposed upon them. A sound that was not gated can now become gated - either trim its length, or have an envelope cut it off.
Almost all envelope tricks deal with the decay or release portion of a sound's envelope. Samplers that allow routing of velocity to decay or release times (humorously the opposite of what keyboard players tend to want - they like to have velocity affect the sound's attack stage) are great at simulating another unusual artefact of struck objects - namely that hitting something harder tends to make it ring out (or cry out, in the case of our kid brother) longer.
This feature (as found on samplers like the Prophet 2000 and Studio 440) is particularly useful for processing long open hi-hat samples by allowing them to vary between an abrupt cutoff (simulating a closed hi-hat) to extending out to their full length, faking many of the nuances of live hi-hat playing. If the sampler you're eyeing does not have this feature, many of the effects can be simulated if it has a "second release" or "hold" function (allowing a footswitch to select between a normal release time and a second or "infinite" time). Failing that, you can always control the sustain stage of the envelope by playing the samples from a keyboard; many drum pad-to-MIDI converters do not allow dynamic variance of the gate time.
TIME TO MOVE on to tricks (and features) that are a bit more esoteric, but certainly useful. Some samplers (like the Prophet 2000, Studio 440, and forthcoming Simmons SDX) provide a feature known as velocity start-point modulation, in which velocity is used to set the point at which playback starts in the sample itself. This can be extremely effective in breaking out of the "same sound, different volume" syndrome, and works best with sounds such as snare drums, which have a rapidly-decaying attack transient.
Sample, or load from disk, a drum that has been hit extremely hard. Next, set the sound up so that lower velocities cause playback to begin further into the sound. When this happens, some of the attack transient is skipped - we come in on the sound at a point where it has already settled down somewhat in both timbre and volume, and this sounds remarkably like a softer hit on the drum (velocity-to-VCA or -VCF may, at this point, seem superfluous). However, some restraint is in order with respect to the amount of this modulation, otherwise you may miss the sound entirely on the softer hits by coming in at the end.
Unfortunately, jumping into a sample at random points like this can create pops and clicks in the attack. These are most noticeable on soft hits, where the attack transient of the sample is not played and is therefore not available to mask them. Notching the VCA attack time up slightly (ie. making it slower) from the usual instantaneous setting used for drum sounds will cover these noises up nicely, but will destroy the hard edge of the attack on hard hits. Unless, that is, the same sampler allows routing of velocity to attack rate - you can then set it so that higher velocities cause the VCA to open more quickly.
This method provides an interesting effect of "distance" modulation, precisely because the softer hits have a less distinct, more diffuse attack - qualities which our ears associate with sounds arriving from a distance away.
Another feature is that of pitch-bend envelopes. Ever notice how a tom tends to drop in pitch from the initial strike? This is because the drum head is stretched from the impact of the blow, raising its pitch. As things calm down, the head relaxes, and the pitch settles back to "normal". The harder you hit the head, the more it is initially distorted, creating a higher initial pitch and a deeper bend. While this effect tends to be over-mimicked by electronic drums, it's useful - or just plain interesting - to impose it on other sounds. And as you've no doubt guessed, the ability to route velocity to the bend depth will more closely mimic the real thing.
This is as good a place as any to mention three other tricks that are possible on virtually all samplers. The first is transposition - very few drum machines allow you to tune their sounds, whereas all samplers allow this. A sampler's pitch range also tends to be as wide as, or wider than, a typical drum machine's. Now you can fine-tune a conga or snare to match what you really had in mind; play chromatically along with the rest of the music; create a set of octabans or a 12-tom kit from a single sample; or just go for weird effects.
Another trick is the ability to reverse a sound - perhaps the perfect reverse reverb, assuming you're not already sick of that sound. Mating reverse playback with gated release time allows precise timing of where that sound ends.
The third trick is looping. To be honest, looping does not work well with the vast majority of percussion sounds - there's just too much difference between how our sounds start and how they finish up (toms included). However, long crossfade loops are useful for those sounds which do have a steady-state period - such as ride cymbals, gongs, and open hi-hats - in which case memory can be saved, and unnaturally long samples can be taken by crossfade-looping; a sampling keyboard generates its longer sounds in this manner. Remember to take a longer sample than you plan to keep, so you have some extra sound to fold back in for the crossfade - you can throw that extra bit away when you're done. And some sort of filter envelope is usually necessary - few things can be as eerie as cymbals that refuse to die...
HAVING COVERED THE setting-up of individual sounds in isolation, it's time now to deal with more than one sound going at once. One of the advantages that samplers have over the average drum machine stems from the flexible manner in which they allow samples to be played back. Often referred to as "voice handlers", these modes of operation are an area of control which simply does not exist on most drum machines, but must be understood in order to make them work for you and to understand and recognise subtle feature differences between different machines.
We'll now cover a sampling (sorry) of issues and answers related to output control, along with tips on how to exploit some of the different voice handler options and keyboard/playback modes.
Typically, you have eight voice channels at your disposal. This is already more than most drum machines have, with the added option of polyphonic assignment. In plain English, this means that a given sound can be triggered many times, and each instance of triggering will be given a different voice channel to play back on, instead of all of them stepping on each other in an attempt to get out on the same channel. Use this with any sound which is prone to being re-triggered before dying away (that is, most of them), and amazingly enough, cymbals fade smoothly into one another, tom fills ring in fullness, and snare drum rolls sound like snare drum rolls, by gum.
Wonderful. But there's a catch. Actually, there are a few catches. In some cases, a mass pile-up of voices on one sound is not what you want. Ever noticed how, on your drum box, the closed hi-hat faithfully cuts off the open hi-hat if it's ringing? This is because both sounds play through the same voice channel. In sampler parlance, this is known as fixed assignment (monophonic playback and so on), and the hi-hat example is one reason why it has been carried over to samplers. A more basic reason applies to machines in which each voice channel appears at a separate output. In order to make effective use of individual audio outputs (for example, to process sounds individually), it is necessary to confine each sound to one of the outputs (and probably also necessary to keep other sounds out of that output).
Wait. We're not done. Not all samplers provide individual voice outputs, and not all of them provide a monophonic assignment mode. For treating one or two sounds differently from the rest, it's nice to be able to use just one or two of the voice channels in monophonic mode and leave the others in dynamic assignment mode. Some machines (like the Emax) will let you do just that; others (like the Akai S900) restrict you to setting the mode for groups of four voices at once; while still others (like the Prophet 2000 and Akai X7000/S700) require all voices to be in the same mode at any one time, limiting their usefulness as total drum sound playback units. The more flexibility your sampler has in using monophonic and polyphonic playback modes simultaneously, the better off you'll be.
There's a hidden gremlin in polyphonic assignment. Hitting one pad very fast, or two pads with the same sound on them, can mean that two identical sounds get started so close together that they actually flange. This can be avoided by having two pads play the same sound at slightly different tunings - say, a semitone apart. This sounds more realistic, particularly with snares, than hitting exactly the same sound - on two-handed rolls, the drummer's one hand is always stronger than the other, or hits the head a little further away from centre, causing the two-hand strokes to sound different.
So much for assignment modes and making sure we don't get burned. Time to start mentioning the potential of some "keyboard" modes samplers provide that make our life more interesting. As we said earlier, note which of these you think you'd like to try, and look for them on the sampler you're considering purchasing.
Perhaps the most common first - a keyboard mode called "velocity switch". What this means is that hits below a certain level play back one sound, while hits above that level play back a different one. Obvious uses include switching between normal- and hard-hit snares, palmed and slapped congas, and so forth. What may not be so obvious is using this feature to switch between two similar but subtly different sounds - such as two different kicks, or two different strokes on a rhythm guitar.
Remember, we're trying to imitate humans, and humans can rarely do the same thing in exactly the same way twice in a row. Tricks like this go a long way towards muting the cry, "I'm a machine".
Next come various ways of layering sounds. A straight layer of two or more samples allows you to build huge, unlikely mixes of sounds - a trick studio engineers use all the time (such as layering four kicks to get one monster bass drum). Many samplers allow velocity to crossfade between two or more sounds - a very soft hit gets one, a medium hit gets both, and a very hard hit gets the second.
And although not quite as versatile, positional crossfading (a low note plays one sample, with higher notes playing a mix including more and more of the second and less of the first) is also good for getting smooth mixes between two sounds over pitch, such as from a high octaban down to a low floor tom.
Other tricks, such as pressure or mod-wheel crossfading between two sounds, are great for keyboard players, but are of little practical value to us - often, all we get is a pitch and how hard it was hit. If you are buying a sampler strictly for percussion, don't let features like these lure you into buying more than you'll need. On the other hand, if you are planning on becoming a drum sound demigod, look for samplers that allow the above layering or switching modes with more than two samples. The Akai X7000, Roland S50, and Casio FZ1 are examples of such samplers. Layering two samples is quite enough for most of us, but taking the time to program more is well worth the more realistic and varied results.
THAT'S OUR STUDY of the playback stage. In the next few instalments, we hope to be looking at initiating those sounds. And in the spirit of the golf course greenkeeper's motto - that you've got to think like gopher to kill gopher - we'll be taking the notion that you've got to swing at things like a drummer or percussionist to sound like one.
Until we "beat" again...
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!