Sample + Synthesis
Most of the current generation of digital synthesizers use a blend of sampling and synthesis techniques to produce their sounds — a powerful combination, but perhaps less easily understood than analogue synthesis. Synth guru Craig Anderton offers some hints and tips on how to get the best out of 'S+S' instruments.
Sample + synthesis (S + S) synthesizers like the D50, M1, Proteus, VFX, DPM3, SY55 and so on, have appropriated sampling technology to give highly realistic imitations of acoustic instruments. Unfortunately, as powerful as these machines are, with a few exceptions — VFX Transwaves, Wavestation wave sequencing, and Yamaha's use of samples as FM synthesis components — the samples are generally short, static sounds. After getting blown away in the store by a fantastic collection of factory presets, disappointment may have set in when you got the synth home and realised that there wasn't a whole lot beyond those presets. For example, maybe the on-board sax sound is great, but how are you going to use it for anything other than creating sax sounds?
If you feel your hard-earned money didn't yield the expected return, or if you're frustrated because you can't afford to upgrade to a more powerful synth, keep reading. A little programming savvy can turn on-board samples into larger-than-life acoustic timbres, impart expressiveness to static sounds, and best of all, create entirely new types of sounds.
The main reason why electronic sounds often aren't as interesting as acoustic sounds is that the latter evolve over time in a very complex way. Synthesizers tend to produce somewhat static sounds that fail to hold our interest after the initial attack transient; sampled sounds generally aren't much better since despite an interesting attack, they often settle down into a repetitive loop. Once you introduce variations in a sample to increase the degree of evolution over time, you'll be well on the way to producing more interesting and life-like sounds.
Don't be put off by rumours that programming a synth to your liking is difficult. Once you've figured out: 1) how to select parameters for adjustment and 2) how to change the parameter values, you have all the knowledge you need. From that point on, it's just a question of experimentation - vary some parameter values and listen to what happens. As you experiment, you'll build up a repertoire of techniques that produce the sounds you like.
The following tips come from working with a variety of sample + synthesis devices. Although these tips are as generic as possible, of course there are differences between different machines, so some adaptations may be required to get these techniques to work with your synth. In a few cases, a tip may simply not apply to your synthesizer. Nonetheless, you can often figure out a workaround, particularly if the synth lets you create layered programs, since you can build up sounds by programming more layers.
We're assuming the synth's architecture is fairly standard: two oscillators per voice ('primary' and 'secondary' voices in Proteus-speak), a filter, output DCA, and modulation sources such as envelopes, LFOs, velocity, and pressure. If your machine has additional goodies, so much the better.
While it may seem that samples must always have a fixed context — eg., an acoustic piano sample can only make acoustic piano sounds — that is not the case. Here are some examples of how samples can be turned into something other than what they were intended to be, along with other oscillator-oriented tricks.
• Try transposing bass samples up by two octaves or more; their characters will change completely. So far I've unearthed good dulcimer, zither, and clavinet sounds by simply transposing bass samples. Furthermore, because transposing up shortens the attack time, bass samples can supply great attack transients for other samples that lack punch. Also, bass samples sometimes make very 'meaty' keyboard sounds when layered with traditional keyboard samples.
• Many S + S devices include 12-string guitar samples, but these are very difficult to loop and, as a result, are often unsatisfying. I prefer to layer two guitar samples, but with one tuned an octave higher than the other. If you can delay the onset of the upper octave signal's envelope a bit (around 10 to 20ms), the effect will be more realistic — although with Rickenbacker guitars, which usually string the octave strings in front of the fundamental strings, you'll want to delay the fundamental sound instead.
• Combining samples with traditional synth waveforms can create a much richer overall effect, as well as cover over problems that may exist in the sample, such as obvious loops. For example, mixing a sawtooth wave with a string section sample gives a richer overall sound (the sawtooth envelope should mimic the amplitude envelope of the strings). Combining triangle waves with nylon string guitars and flutes also works well. And to turn a sax patch into a sax section, mix in some sawtooth wave set for a bit of an attack time, and detune it compared to the main sax.
• Transpose percussion sounds way down (two or more octaves) for weird sound effects and digital noises. Since there will probably be a lot of quantisation noise and grunge, you may want to close the filter down a bit. For massive thunder and spaceship sounds, choose a complex waveform, transpose it down as far as it will go, and close the filter way down.
• When creating an unpitched sound, such as drums or special effects, use two versions of the same sample for the two oscillators, but with their pitches offset by several semitones to thicken the sound.
• If you're into 17-tone, 21-tone, or other exotic even-tempered scales — or simply want a 1-tone scale where all notes on the keyboard play at the same pitch — assign note position (keyboard) as an oscillator modulation source. Adjusting the degree of modulation can 'stretch' or 'compress' the keyboard so that an octave takes up more or less keys than the usual 12. Note that you will need to adjust the tuning so that the 'base' key of a scale falls where you want it to fall.
• Sometimes, combining theoretically dissimilar samples works well. For example, on one unit I noticed the piano sample lacked a strong bottom end. Layering an acoustic bass sample way in the background solved the problem. Adding a sine wave fundamental to a sound can also sometimes increase its depth; this worked well with a Chapman Stick sample to increase the low end.
Unexpected combinations can sometimes yield useful results, such as mixing choir and bell samples together, or high-pitched white noise and choir.
• Sample start point changes can radically affect the timbre and add dynamics (unfortunately, not all S + S synths can do this). Move the start point further into the sample until you obtain the desired 'minimum dynamics' sound, then tie the start point time to keyboard dynamics (velocity) so that more velocity moves the start point closer to the beginning of the sample (this usually requires negative modulation, but check your manual). This seems to work best with percussive sounds, since changing the start point dynamically can cause clicks that are particularly obvious with sustained sounds. An alternative is to use two versions of the same sample, with one sample's start time set into the sample and the other left alone, then use velocity switching to switch from the altered sample to the unaltered one as velocity increases.
• Use waveform crossfading to cover up samples with bad loops. For example, one unit had a great flute sample attack, but it was stretched across the entire keyboard. Playing high up on the keyboard gave a short, tinny loop that negated the excellence of the attack. So, I used the flute for one oscillator and a triangle wave for the other, and faded out the flute while fading in the triangle wave (Figure 1). The flute sample provided the attack, and the triangle wave a smooth and consistent post-attack sound. Similar techniques work well for brass, but you'll probably want to crossfade with a sawtooth wave or other complex waveform.
• Layering two samples, and keying one sample's amplitude to velocity so that hitting the keys harder brings in the second sample, can be very effective in creating more complex sounds. For the second sample, try a detuned version of the original sample, so that playing harder brings in a chorusing effect; another useful option is to bring in a version of the original sample tuned up an octave, or an entirely different sound altogether.
• When introducing LFO vibrato on big pads (orchestras, strings, etc.), be creative. If two different LFOs are available, send one to each sample, with each LFO set to a slightly different speed, and possibly with negative and positive modulation. This will create a more animated sound. Even if you have only one LFO, at least try sending positive and negative modulation at varying levels to the samples. Many of today's synths aren't fast enough to process both waveforms at exactly the same time, giving slight variations.
• If your synth doesn't offer LFO-controlled panning, sometimes you can simulate the effect by creating two layers and panning them to opposite sides of the stereo field. Then, set up out-of-phase LFO amplitude modulation for each layer (ie. as one layer gets louder, the other becomes softer). For example, the Yamaha SY55 has a phase control that lets you do this; with other synths, apply positive and negative LFO modulation to create the panning effects. For a more randomised panning effect, use two different LFOs for each layer's modulation. Set them slightly out of sync for the widest range of variations.
• If you can modulate LFO depth, or if your synth has modulatable levels for each sample as well as a modulatable master level, LFOs can provide primitive echo effects. Use a downward ramp (negative-going sawtooth) for traditional echo effects, or square wave modulation to 'chop' the signal. Set the LFO for a fairly slow rate, and modulate the amplitude of the sample(s) (the LFO should be the sole means of sample amplitude control at this stage). If you can modulate the LFO level, do so with a decaying envelope so that as the note decays, successive 'echoes' become quieter (Figure 2). With synths that have modulatable levels for each sample, and also for the master level, use an envelope to control the master level so that each 'echo' gets quieter as the note decays.
• LFOs can also serve as extra attack or decay envelope generators. Use a slow ramp up or ramp down signal, which will usually be triggered when you hit a key, to control the parameter to be 'enveloped'. As long as the notes you're going to play end before the LFO reaches its next cycle, you're okay.
Many S + S synths use rate/level envelopes, where the envelope shape is defined by several levels, and the time it takes to get from one level to another (Figure 3). In most of the examples below, we'll assume five levels and four times. Many of the following tricks are not as well, suited for envelopes that use the more traditional ADSR structure (Figure 4) but they still translate in some cases.
• For sounds whose attacks are too prominent, you can tame them by dropping the first level below the second level (Figure 5), and setting the first time equal to the time of the attack to be tamed (in the example shown, the other levels are set to produce a sustaining sound but, could just as easily give a decay, second attack, etc.).
• A variation on this theme is to 'compress' a percussive waveform by programming the first level or two to create an amplitude envelope equal but opposite to the sample's natural envelope. This works similarly to the example in Figure 5.
• Tying attack time to velocity so that higher velocities give shorter attacks can lead to more expressive wind and string parts. Try this on flutes and trumpets so that you can articulate the note attacks more expressively; with low velocity playing, you won't hear the characteristic 'spit' or 'chiff' at the beginning of the sound.
• Even if your envelope generator doesn't have an adjustable attack delay before it kicks in (good for echo/delay effects and double attacks), most rate/level envelopes can be programmed to create a delay. Set the first and second level to 0; the first time setting determines the delay before the onset of the note (Figure 6). If you release the key before the delay time is over, the note will not play.
• Modulate the filter cutoff with velocity for more realistic dynamic control. Programming velocity to open up the filter somewhat will make the sound brighter as you play harder, which increases the apparent dynamics. Filter envelopes whose levels can be modulated by velocity provide similar effects.
• Filters can serve as primitive tone controls and exciters if there's a resonance parameter. Leave the filter cutoff at a static (non-modulated) setting and turn up the resonance to introduce a peak in the sound. Combining a high cutoff setting with moderate resonance will provide an 'exciter' type of effect by boosting the treble.
• Sometimes samples become shrill-sounding when you bend pitch up because of sample rate transposition, especially if the pitch bend range is wide. Try tying negative pitch bend modulation to the filter cutoff so that bending pitch up lowers the filter cutoff.
• Most units let you create splits from different patches (called combis, multis, etc.). On power chord/lead guitar splits, I usually set the mod wheel to bend pitch down on the power chords (for whammy bar effects), and the pitch bend wheel to bend the solo voice's pitch. This gives independent bend control over each part of the split.
• Voice and wind instruments often do not start out right on pitch, but begin a little flat and slide up to pitch. To simulate this, program a very slight upward pitch bend at the beginning of the sound. For even better results, use the pitch bend wheel to add these changes dynamically as you play.
• Try adding a very small amount of delayed random modulation to trumpet and other brass sounds, which are difficult to keep on a constant pitch. Random modulation can also help loops sound less static. Use very little modulation (a little randomness goes a long way!) and delay the onset of modulation until the loop appears.
• Random detuning used to be a 'feature' of every analogue synth, as oscillators drifted and went out of tune. In many cases, this was responsible for the 'fat' sound associated with analogue synths. Digital synths put an end to the drift problem, but Yamaha added a random detuning feature to their instruments, starting with the DX7 II, in order to imitate this effect. If your keyboard doesn't have random detuning, set each oscillator for a small amount of complementary detuning (eg., +1 and -1 cents) and subtly modulate the pitch with velocity or pressure. As you play with varying degrees of force, the degree of detuning, and hence the 'chorusing', will change so that no two successive notes sound exactly alike (unless you manage to hit them both with identical dynamics, which is unlikely).
• Using the mod wheel solely to add vibrato doesn't take full advantage of what a mod wheel can do. For example, with bass patches, use the mod wheel to control the level of a secondary sample set an octave lower. As you rotate the mod wheel, you'll be able to bring in a sub-octave sound to add emphasis and increase the sound's hugeness factor.
• Tying the mod wheel to filter cutoff lets you use the wheel as a tone control. I recommend negative modulation, where rotating the wheel away from you lowers the filter cutoff. This is because when you switch from another patch to this one, odds are the mod wheel will be rotated all the way toward you; I prefer having the default version of the patch appear in its full-bandwidth splendour.
• Mod wheel and keyboard pressure can provide chorus depth control, even if your synth doesn't have an on-board chorus effect as such. Program the mod wheel so that it modulates the oscillator pitches and detunes them in opposite directions — ie. one goes sharp as the other goes flat. As with so many of these examples only a little modulation is needed to give a useful effect.
• On lead guitar patches, tune one lead sample an octave higher than the other lead sample and tie both sample levels to keyboard pressure. However, the initial volume of the main sample should be at maximum level, with pressure adding negative modulation that lowers the level; the octave-higher sample should start at minimum level, with pressure adding positive modulation that increases the level. Pressing down on the key during a sustaining note brings in the octave higher 'feedback' sound and fades outs the fundamental. For a couple of variations on this theme, have pressure introduce vibrato and perhaps bend pitch up a half-tone at maximum pressure. Also experiment with alternative waveforms and pitches for the octave-higher sound; a sine wave tuned an octave and a fifth above the fundamental gives a very convincing feedback effect.
• Pressure is great for introducing vibrato, since you can add some without having to take your fingers off the keys and fiddle with a mod wheel.
• Sometimes you want sustained sounds to remain in a mix, but become less obtrusive. This can be done with ways other than lowering the level. With pads and sustained bass sounds, have pressure close down the filter a bit. When you first hit the keys, you'll hear the instrument's full-bandwidth attack. As the instrument continues to sustain, pressing on the keys will mute the sound more and place it more in the background.
• Conversely, sometimes pressure is needed to make a sound more prominent. Although you could just swell the overall volume level, increasing the filter cutoff with pressure can yield excellent results. This is particularly true when you want to 'swell' brass and wind programs.
• For percussive bass sounds where the decay time is set by an envelope's release parameter, you can add some 'punch' at the attack by introducing a little pressure-controlled upward pitch modulation to one sample in a pair of bass samples (either the same or different samples). With percussive sounds, the object is to tap the keyboard keys rather than press and hold them. So, if you tap the key hard enough to trigger the pressure response, there will be a very brief up-to-down pitch change for one of the samples, which gives a sort of slap bass effect. This requires a bit of keyboard playing control, but increases the overall dynamics.
• Using pressure with drums may sound like a dumb idea, but it's very useful for bringing drum rolls in and out, as well as producing bend pitch on tympani patches.
Wave sequencing is the process of having one sample or wave evolve into another sample or wave over time to produce a complex, highly animated sound. Although instruments not designed for this task can't even get close to the effects associated with something like the Korg Wavestation, it is a testimony to the usefulness of this technique that even two-stage wave sequencing, where one sound crossfades into another, makes some great sounds. Fortunately, many keyboards can string several programs together into a 'multi' or 'combi' program where each program can be delayed in time. This lets you program several individual two-stage programs, then trigger each program at an appropriate time so that one 'wave sequenced' program segues into another 'wave sequenced' program, giving four stages total (or more, if your synth can trigger several programs in series).
The basic idea is to set up two crossfading level envelopes for two different samples, similar to the technique shown in Figure 1. Percussive sounds often work well for the first sample. When you press a key, the first sound of the sequence runs its course and crossfades into the second sound. This second sound can either sustain if you want to stick with a two-stage wave sequence, or decay while yet another wave sequence fades in.
Figure 7 shows how you would set up a VFX-family synth to do 6-stage wave sequencing. The VFX can layer up to six different sounds on one key, and each voice has a delay time (selectable on the wave page, with a 'master delay' multiplier on the program control page). By setting attack/decay envelopes and adjusting the delay times properly, you can obtain a highly animated sound.
Well, that should be enough to get you started. Remember that the whole point of programming is to come up with your own great sounds, so hopefully these ideas will serve as a springboard from which you can create your own variations and permutations. Synth programming is a lot of fun once you get into it, but most importantly, it can help give you a signature sound that is as uniquely yours as a fingerprint. Have fun!
Feature by Craig Anderton
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!