Ever since the invention of the log drum and the bone whistle (at a guess, 50,000 years ago), music has been increasingly dependent on technology. But music has never been about technology. It's about expressing feelings. The trouble is, sometimes technology gets in the way of expression. Nowhere is that a greater danger than when a computer gets into the act. In this article we'll explore some ways to avoid monotony and add more expressive colors to your computer-based music.
Technically, you could pull off most of these ideas by drawing controller shapes into sequencer tracks with a mouse (see Figure 1). But ultimately, playing music is a physical activity. If you have access to a decent MIDI controller device with a few knobs and sliders, as discussed in my previous article, "Choosing and Using MIDI Controllers," you'll find it easier to let your intuition take over. That's bound to deepen the expressive side of the music.
To start with, we'll look at some of the sonic changes MIDI controller data is most often used for. Then we'll look at how controller data can be edited in a sequencer. If you have a decent synthesizer, either software or hardware, don't just download the MP3s. Instead, try out some of these ideas as you read. There's no substitute for hands-on experimentation.
Keyboard players routinely use MIDI velocity to make their playing more expressive. Without velocity, for example, a sampled grand piano would sound horribly stiff and rigid, because each note would have the same volume and brightness. (On better synthesizers, velocity controls even more factors—triggering a different sample, for instance.) But velocity is defined only at the beginning of each new note, and can't change during the course of the note. In this article we'll be looking at ways to use controllers to shape single notes (and also phrases) in ways that velocity simply can't accomplish.
In the MIDI world, people use the word "controller" in two ways: to describe hardware such as a joystick or wheel that controls a synthesizer's sound, and to describe the data (technically called a Control Change message) that the hardware transmits. The term controller is also applied to an entire device; you'll hear "keyboard controller" or "wind controller" used to describe hardware that triggers and shapes the sound on a connected synthesizer but makes no sound of its own.
In the data realm, the MIDI specification defines 128 Control Change (CC) numbers, each with 128 possible values (0-127). The most common types are CC7 (volume), CC1 (modulation), and CC10 (panning). Several CCs are defined as switches. CC64, for example, is the damper pedal control. When a synthesizer receives a CC64 command with a value greater than 63, it will usually sustain a note; values of 63 and less will allow the release of any notes being held by the damper pedal. (I say "usually" because this command has no effect on non-sustaining sounds like drums, and can often be mapped to control other parameters instead.)
Pressure (aftertouch) and pitch-bend, two other types of performance data, aren't technically Control Change messages but behave similarly.
Because the human voice has neither frets nor keys, it can do all sorts of expressive things with pitch. The voice can glide smoothly from one pitch to another, "shade" pitches up or down to give them color and emphasis, or wobble around a pitch rather than remaining static. Good singers use these effects every day. And because the human voice is our oldest musical instrument, pitch changes of this sort have an immediate emotional impact on listeners. When we hear the pitch move, we respond.
Not surprisingly, the two controllers most often used with synthesizers both affect pitch. (See Figure 2.) The pitch-bend wheel, lever, or joystick is used for bending the pitch up or down from its expected value. And the modulation (mod) wheel is most often used for adding vibrato, which is a regular up and down wobble in the pitch. On some home keyboards, that's the only thing the mod wheel can do, but in pro-oriented synths it has many other possible functions, like changing tone color or volume.
The maximum depth of a pitch-bend is set (usually in semitones) in the receiving synthesizer. A semitone, or half step, corresponds to the pitch change between two adjacent notes on a piano. Bends of 24 semitones (two octaves) or more are allowed in many synths, but such a huge range is useful only for an occasional "dive-bomb" effect with bass or lead guitar sounds. You'll find it easiest to develop an expressive pitch-bend technique if you always set the bend depth to ±two half steps. That will allow you to bend up or down by a single scale-step in normal music, without having to worry about overshooting. Some advanced players, however, like to set pitch-bend range to three semitones to increase the drama; if you bend a tiny bit sharp or flat from the expected note, it can sound more exciting.
To learn how pitch-bends can be used, listen to a few good vocalists, blues guitarists, and saxophone players. You'll hear several devices used again and again:
A quick scoop up from below at the beginning of a note.
A longer, slow swoop up from below for a high note that climaxes a phrase.
A fall-off at the end of a held note. (This sounds especially cool when you're feeding the synthesizer through a delay line with some feedback.)
A gliding movement between the minor third and major third of the scale, or between the fourth and flat fifth. These "blue notes" are characteristic of the blues.
I've illustrated three of these techniques in the following blues phrase: quick scoops up at the beginnings of a few notes, blue notes bending down from the flat 5th and up from the flat 3rd, and a falloff at the end of the last note. Filter cutoff modulation brings out the long notes, and I add some subtle distortion at the end of the phrase. I used Camel Audio Cameleon 5000 for the lead sound and Wizoo Hypersonic for the bass and drums.
Learning to perform these bends on a synthesizer requires a little practice. The exact nature of the technique will depend on the type of controller hardware you have, but it's a good idea to develop a consistent and comfortable hand position. Try to anchor the side or heel of your hand on the instrument's chassis. Assuming you've set the pitch-bend depth to a whole step, learning to bend a half step reliably is also useful.
If you can bend a half step and then introduce vibrato manually, you're ready to teach a class at Berklee. Most pitch-bend hardware can't be used easily for vibrato because there's a dead area in the center of the hardware's travel. The exception is the pitch stick on Clavia's Nord Lead line of hardware synths (see Figure 3), which responds very well to vibrato movements.
Adding vibrato with the mod wheel is easier than doing pitch-bends; just push the mod wheel, and there it is. If you're creating your own synthesizer sounds, though, pay close attention to the vibrato speed, which is programmed using the low frequency oscillator (LFO) rate control. If it's too fast, the vibrato will sound nervous and twittery. Too slow, and it will sound seasick.
Of course, both nervous and seasick can be useful musical effects. This simple lead—played on Applied Acoustics Systems (AAS) UltraAnalog software synth, with accompaniment from Hypersonic—illustrates two concepts: blending a second, detuned oscillator at the end of the first phrase and through the second phrase, and vibrato that ranges from too slow to too fast:
If you listen to singers, you'll find that they don't always add vibrato at the beginning of long notes. Quite often the vibrato swells during the course of the note, an effect you can easily achieve by pushing the mod wheel forward slowly. Some notes may get no vibrato, while other notes in the same phrase get a little added vibrato for emphasis.
Also, a singer's vibrato doesn't always move up and down at the same speed. Increasing the vibrato speed at the same time you increase its depth will add intensity to a sustained note. Setting up this effect on a synth is not quite as easy, because what you want is just a slight increase in vibrato speed, not a radical change. If your instrument allows you to change two different parameters (in this case, LFO amount and LFO rate) from a single MIDI controller, experiment with the settings to see what sounds most natural to you. For details on configuring this setup, see the "How Much Is Enough?" sidebar.
The simplest way for a manufacturer to implement MIDI control of parameters is to assign the minimum value of a MIDI Control Change message (zero) to the minimum value of the parameter and the maximum value of the Control Change message (127) to the maximum value of the parameter. If you're modulating filter cutoff, for instance, the minimum setting for the parameter might be 20Hz and the maximum 20,000Hz.
But what if you only want a small change in cutoff in response to your controller move? Sure, you can nudge the controller hardware only slightly, but making small moves with precision is hard.
Fortunately, some synthesizers let you scale the depth of the controller response for each parameter. In the example above, you might be able to set the minimum cutoff (when the controller value is at zero) to 750Hz and the maximum value (when the controller is at 127) to 1,200Hz. Now you can grab the slider and pull or push it, and the filter will respond in a subtle way. This type of controller scaling is an extremely useful feature and is found in one form or another on most high-end synthesizers. You may even be able to invert the controller response so that a value of 127 maps to 750Hz and a value of zero maps to 1,200Hz.
Why would you want to do that? Because the synth may also allow you to assign several parameters to a single MIDI controller. Some of the most expressive effects are achieved by lowering one value while raising another. For instance, you might want to lower the cutoff while increasing the amount of distortion. That would let you achieve a thick growl that isn't overpoweringly bright.
If you listen to singers and horn players, you'll find that at moments of peak excitement they not only get louder and add more vibrato, their tone also gets thicker, brighter, and sometimes raspier. A pop vocalist may literally scream a key word or phrase.
Synthesizers are pretty good at mimicking all of these effects. Try these approaches:
Crank it. Your synth may already map CC 11 (the so-called "expression" controller) to loudness. If it doesn't, assign the MIDI controller number of your choice to output level.
Brighten it. By assigning a controller to the cutoff frequency of a lowpass filter, you can make the tone swell, adding brightness as needed. We heard this in the pitch-bend example (300KB MP3).
Drive it. Increasing the input gain or drive on a distortion effect will add raspiness, dirt, and growl.
Double it. Another way to increase the animation in the tone is to blend in a second, detuned oscillator under MIDI control, as we heard in the seasick vibrato example (332KB MP3).
These types of control are good candidates for aftertouch, a MIDI controller message that's generated by pressing down on the key after it has reached the bottom of its travel. (See the "Touch Up" sidebar.) Unfortunately, not all MIDI keyboards transmit aftertouch. If yours doesn't, you should be able to assign some other available physical controller, such as a slider or joystick, to control filter cutoff or distortion drive amount.
Ignorant or careless writers sometimes create confusion by referring to key velocity as "pressure." But actual MIDI pressure messages, also known as aftertouch, can be sent continuously during the course of a single note. (I've also seen writers refer to velocity as "the speed of the attack," which is very confusing.) Manufacturers who don't want to confuse the Great Unwashed with techical terms like like "velocity" and "pressure" use the term "touch-sensitive," but unfortunately this term is ambiguous. It can refer either to velocity sensing (which most MIDI keyboards have) or to aftertouch sensing, which is less common in the consumer world.
The mechanism keyboards use to sense pressure varies as well. On some keyboards, there's quite a bit of squishiness when you press harder on a key that's already resting on the keybed. Others barely move at all. The majority of keyboards with aftertouch use what's called channel pressure, meaning that the key that's pressed the hardest will determine the pressure value and affect all notes sounding on that MIDI channel. A handful of keyboards support the more expressive key pressure, which transmits a separate pressure value to each note.
Most synthesizers have dozens or even hundreds of parameters, any of which are fair game for expressive changes. Modulating (changing) most of these parameters won't emulate anything a singer or horn player would ever do, but so what? The point is to express yourself, not to slavishly imitate someone else. Here are some ideas to try:
Attack it. By modulating envelope attack time, you can give some notes a soft, rounded attack and others a snappy, instant attack, perfect for articulating lines that include both fast notes and long notes.
This short figure (played on AAS UltraAnalog) uses four different controllers. CC1 adds vibrato. CC2 speeds up the vibrato, most noticeably during the last note. CC3 affects envelope attack time and sustain level together, so that low controller levels produce quick attacks and notes that decay to a low level, while high controller levels produce long attacks and notes that sustain at a high level. CC6 speeds up two LFOs, which are modulating both filter cutoff and panning, with the result that the sound swirls (during the long notes in the middle of the phrase) in a vaguely Leslie-like way:
The melody in the previous example sounded a little silly by itself, so I recorded an accompaniment using Wizoo Darbuka for percussion and Hypersonic for bass. Some of the modulation is harder to hear in this file, but the musical intent is clearer:
Wet it. With a reverb or delay line, increasing the amount of an effect return (perhaps by modulating the wet/dry mix parameter) will allow you to create a ghostly halo at the end of a phrase without smothering the middle of the phrase in the effect.
Pan it. Try assigning an LFO to control stereo panning (or use a pair of LFOs for quad panning, if you have 5.1 or quadraphonic surround outputs). Then use MIDI to modulate the rate of the LFO from slow to fast. Slow panning can be almost dreamy, while fast panning can be dizzying or frenzied. (The Phrygian Attack solo illustrated this.)
Shape it. You can add interesting changes in timbre by modulating waveshape, the cutoff of a comb filter, or the depth of audio-rate frequency modulation (FM).
All of the types of control described above, and many others we haven't mentioned, can be done in live performance. In the studio, though, you may want to take advantage of your computer sequencer's MIDI editing and overdubbing features to shape your controller moves more precisely.
Some people prefer to record the notes of a MIDI part first and then add controller data as overdubs. I find that this works pretty well with modulation, whether I'm adding vibrato with the mod wheel or making some other type of sonic change. But I prefer to play pitch-bends at the same time I'm recording the notes, because the timing relationship between the bend and the beginning of the note is so important.
When you examine the data being transmitted by your controller hardware, you may find that it skips over certain values, producing a rough staircase shape, as shown in Figure 4. If the controller is increasing the amount of a vibrato LFO, the stairstepping of the data probably won't matter, because the LFO modulation will still be smooth. But with pitch-bends, you may want to smooth out the data using a pencil tool, as shown back in Figure 1 (52KB JPEG).
If you're modulating the cutoff frequency or panning of an electronic-sounding track (such as a repeating loop) you may want to enter the data directly in the sequencer rather than trying to play it live. This can be done using the graphic straight-line tool in your sequencer (see Figure 5).
Another useful feature of sequencers is the ability to scale controller data. The details of doing this differ from one sequencer to another, but the basic idea is to select a bunch of MIDI events and then multiply the value of a specific parameter by some amount. Most often, I'll use scaling if I've been a little too energetic with the mod wheel—multiplying all values by 0.5 or 0.8 to reduce the height of the controller contour without changing its shape. For technical reasons, some sequencers don't do a good job with scaling pitch-bend data, so your best bet may be to erase the pitch-bend and rerecord it.
Quantizing (time-correcting) the rhythms of notes can sometimes have an undesirable effect on the sound of controller data. Only in the last couple of years have we started to see sequencers that will intelligently shift controller data forward or backward in time when the note above the controller data is moved. If your sequencer does this, your pitch-bends and other moves should remain aligned with the notes after quantizing.
If quantizing destroys the relationships between notes and other data, you have three options: don't quantize the notes, "quantize" by hand, dragging each note into a better rhythmic alignment and then selecting and dragging its controller data by the same amount, or re-record the part to get better timing. This is a lot of work, but the results will probably be worth the extra effort.
The natural world is full of curves, from the track of a snail or the arc of a tree limb to the bays and promontories of a seacoast. Adding a few controller curves—or a lot of them—will help your music seem more organic, less mechanical. And in an overly rigid, mechanized world, that's surely a good thing.
Return to digitalmedia.oreilly.com