Using Onboard Synth Effects, Part 4

Part 4: In the conclusion of his series on DSP effects built into synthesizers, Paul Wiffen looks at those effects which rather than delicately colouring a sound, transform it beyond all recognition.

Preferring to end with a bang rather than a whimper, I have saved for the final instalment of this series a discussion of those effects which cause a Jekyll-to-Hyde style transformation in the sounds which you put though them. My first introduction to effects of this sort came not on a synthesizer at all, but on the electric guitar. A friend in the school science club introduced me to the Practical Electronics 'fuzz-box-in-a-tobacco-tin' phenomenon -- which, to my ears, was nothing short of a miracle. Take a handful of cheap components and shove through them the thin plinking that the pickups on a cheap electric guitar produce and suddenly you had a sound with more harmonic content, sustain and interest than had even been in control of before.

The digital descendants of such primitive amateur projects may be much more refined (and reliable!), but they share something fundamental with their precursors: they bring more to the final result than whatever sound is fed through them. The fundamental reason for this is that whilst the effects we looked at last month take the frequencies present in the source signal, replicate them and combine them in ways which subtely enhance or reinforce them, the effects under scrutiny here actually change the source signal so much that they introduce completely new frequencies and overtones which can introduce character and interest that the source sound may not have had.

The way that distortion produces these additional frequencies is by completely reshaping the source waveform. When a signal is overamplified, it is not just the loudest points in the original which end up at maximum amplitude in the result. Much of the original waveform is pushed up to the maximum levels, resulting in a complete change to the resulting shape. In fact the shape of the louder points of the waveform is lost, due to their being flattened along the maximum output limits -- an effect known as clipping, because the resulting waveform looks as though all its most prominent peaks and troughs have been clipped off (see right).

This is where the additional harmonic content in an overamplified signal comes from. The simplest sonic signal, a sine wave which only contains the fundamental frequency, is a smoothly and constantly changing shape. Square or pulse waves, on the other hand, contain lots of higher harmonics in significant proportions to the fundamentals, and spend the majority of their cycle at maximum amplitude. This suggests that if something artificial is done to increase the amount of time a source waveform spends at maximum amplitude, the more complex its harmonic content will become.

Now, if you overdrive a signal which already has lots of harmonic content (like a complete mixed recording), the result tends to be unpleasant by anybody's reckoning, as the already complex content is rendered a complete mess of frequencies. So when we distort a hi-fi amp by overamplifying the stereo signal fed through it, we end up with an unlistenable sound because there was too much in the original signal.

Clearly, then, when this effect is made available on a synth, it is not for processing the entire output of your multitimbral setup, or providing a refined finishing touch to a sound. If you take distortion (or overdrive, or whatever it is called on your synth) off a factory program which uses it, the 'sound' is usually completely lost and the result rarely has any value in itself. It is therefore the sort of effect you should reach for when the sound you want to apply it to is insipid, weedy or just lacking in 'bite'. Some of the best sources to apply it to are those single-cycle PCM waveforms included by most S+S synth manufacturers, which in themselves don't sound very interesting.

Of course, you don't have to use distortion/overdrive exclusively on thin, uninteresting sounds. Occasionally you may find that adding a little distortion to the wet part of an insert-effected sound gives you a distant hint of a more menacing relative of the clean sound, which can help to make the atmosphere much darker (especially if the upfront clean sound is just too 'nice').

One major reason why the distortion effect caught on with guitarists is that it acts as a form of compression, which results in increased sustain. A 'straight' guitar sound has a sharp attack transient which dies away fairly quickly to a sine wave (the fundamental). Not only does overdrive increase the sonic interest in this rather simple waveform, but it also compresses the sound, limiting the peak of the attack and then raising the level of the signal as it tails off.

If you find yourself with a PCM sample of limited length or with a quieter looped section, you can use this aspect of overdrive or distortion to your advantage. As well as changing the harmonic characteristics, the overdrive will tend to smooth out any sharp transient at the beginning and bring up the tail end of the sound. So if your source sound is not only feeble but tails off rather too quickly, then distortion may fix both problems at once. Of course, one aspect of distortion is that it tends to make anything sound like a distorted electric guitar -- but keep the levels down and some of the original character of your source will come through.

There are, of course, other effects present in the DSP effects sections of modern keyboards which trace their ancestry back to the guitar stomp boxes and pedals of yesteryear (wah-wah, which is basically a foot-controlled band-pass filter, is the first example that springs to mind). Some of these can change a sound as radically as distortions and overdrives (even to the point of adding a vocal element with effects like Talk Mod), and all are as worthy of investigation and experimentation. However, if we tried to cover every effect now available on a modern keyboard, this series would run for most of the next millennium. I will, therefore, end this look at built-in effects on synths with an examination of a relatively new type of DSP algorithm which, although I hear or read very little about it elsewhere, seems to me one of the most exciting.

Although samplers these days pride themselves on the fidelity of their reproduction, it wasn't always like that. There was a time when even 16-bit/44.1kHz sampling was regarded as science fiction. The first commercially available sampling systems (even the cripplingly expensive ones like the Fairlight and Emulator) offered 8-bit resolution and sampling rates as low as 8kHz (although this soon increased to a massive 16kHz).

When you make samples with these bit depths and sample rates, the results can hardly be described as transparent. Various artifacts can appear in the samples, including a harsh crunchiness caused by the low bit depths, and a metallic rattling noise in the upper frequencies known as aliasing, a by-product of the sampling frequency's interaction with frequencies around half the sample rate -- see the 'All About Aliasing' box.

A sample rate of 44.1kHz is high enough that input frequencies likely to cause aliasing will be above the audible range (since half the sampling frequency is 22.05kHz), and can therefore be eliminated before sampling using a steep low-pass filter on the input, without adversely affecting the reproduction of the source sound. Back in the olden days, however, creative souls found that they liked the way sounds changed when sampled at lower frequencies, even when aliasing artifacts were audible. Peter Gabriel, who was looking for elements to add unease or even menace to his music, made great play of the Fairlight's ability to turn even the nicest source into a nasty dark sound. Others, like The Art Of Noise, found that samples made at lower rates and resolution stood out in a track by virtue of their jagged crunchy nature.

Just as the high-end professional samplers began to get high enough sample rates and resolution to eliminate audible sampling artifacts, a new wave of cheaper samplers led by the Ensoniq Mirage and Emulator SP12 brought lower-fidelity sampling to a whole new generation of users. Many of these 8-bit samplers fell into the hands of experimental souls less concerned with traditional musicality and more interested in radical new sounds which the aliasing and low bit resolutions provided. Early rap music, for example, features drum sounds so crude that the source instrument is unrecognisable, but they give a distinctive grungy edge to the music. With the advent of industrial and techno music, these grungy samplers found other aficionados (aliasing tends to make the most musical source sound as if it being produced by a piece of industrial machinery) and the use of aliased, grunged-up samples proliferated.

Those of you working on musical forms which favour this downgraded quality will be pleased to know that it is no longer necessary to acquire these older samplers to be able to achieve this type of sound. Several state-of-the-art modern keyboards now feature DSP effects which can do precisely the same job on any source sample in real time. It is a fairly simple process to remove some of the information contained in modern high-resolution PCM samples -- certainly less demanding of DSP power than effects like reverb, which require adding extra content to the signal. This 'downgrading' may take two forms: reducing the sample rate and/or the bit resolution.

As you can hear from the samplers of the early '80s, reduced bit resolution tends to introduce a crunchier, more upfront sound (again, compression appears as a by-product of the process, as the dynamic range available is reduced). This can be very useful for making elements in your music (especially percussion elements or lead parts) really stand out in a track. It is particularly effective on percussion sounds, so if you find the drum sounds provided in your S+S keyboard too nice and clean for your depraved tastes, try running them through bit reduction for that authentic mid-'80s crunchiness. They will become much punchier and more aggressive, and crash through whatever else is in your track.

Try running one or more elements of your track through bit reduction and then steadily reducing the bit depth. As you take more and more bits out of the signal, it will get nastier and nastier, till the sound comes in jagged lumps which finally fall apart completely. I have used this technique on a repeated riff to avoid the cliché of fading out at the end of a tune, producing a much more striking complete disintegration of the track. So far, however, I have not managed to find an

When it comes to sample-rate reduction, however, several modern machines will let you use a real-time controller to let you play around with the effect. On the Korg Z1, for instance, you can use the X-Y Pad to control the sample-replay rate of their Decimator in tandem with another parameter. It is only when you start to control radical DSP effects like sample-rate reduction in real time that you realise the extent of their full potential.

It is very interesting to hear exactly what happens to a signal as the sample rate is slowly reduced. Obviously, the faithful reproduction of the top end is the first thing to suffer. The higher frequencies will become sharper and more grainy to begin with, and then as this affects progressively lower frequencies, the audible top end becomes more and more disassociated from what that of the source as the unrelated aliasing frequencies take over. The sound becomes more and more of a caricature of the original, containing the same basic elements but sketched more and more roughly and inaccurately. In the end, even the fundamental basis of the sound is transformed into something else, the pitch of which is more related to the sample rate and related aliasing frequencies than to the pitch of the source sound. When pushed to the extremes of, say, a 4kHz sample rate, aliasing becomes so extreme that even percussion and other unpitched sounds acquire a definite pitch and sound more like a bell than a drum. On one track I actually used this pitched effect on a rhythm pattern to start the track as if it were a pitched sequence, and then slowly increased the sample rate until the percussion elements lost the pitching and 'cleared' into the rhythmic basis of the track, at which point I then added 'real' pitched parts. Few effects allow you to change the nature of sound as radically as monkeying with the bit depth and sample rate. It probably won't fit too well into a ballad or even an standard pop song, but if you are looking to give your music an aggressive, evil edge, I can thoroughly recommend it.

These total transformations may be just what you want, but just in case they aren't, some sample-rate reduction effects like Korg's Decimator feature an integrated low-pass filter to allow you to remove some or all of the frequencies that cause aliasing. The function of this filter is to allow you to achieve really significant rate reductions without the source sound being completely obliterated by aliasing. This means you can hear the more subtle effects of sample-rate reduction which would otherwise be obscured by the proliferation of unrelated aliasing frequencies.

The most obvious of these effects is that the sound becomes duller (as you would expect if a low-pass filter is being applied), but there is more to it than this. The sound also becomes brasher and more 'rock & roll', even Neanderthal. Presumably this is why Fairlight and Mirage aficionados still say that their instruments of choice compete better with raunchy guitars and raw drum sounds than modern-day samplers.

The very radical nature of the effects I've been looking at in this instalment can produce results which not only defy description, but can take your music in very different directions. For the last couple of years, I have been experimenting with effects like these, and find that they are the perfect antidote for the super-clean, super-quiet, often bland sound which the transparentness of modern digital technology can lead to in S+S keyboards. The real joy of the DSP facilities we have looked at in this final segment is that now that the characteristics they can produce are no longer endemic to the music-making process (forcing musicians to live with their effects or work around them, as was the case when the technology was evolving) -- you can choose to put as much or as little of these rawer elements in as you want.

Home | Search | News | Current Issue | Digital Editions | Articles | Forum | Subscribe | Shop | Readers Ads

Advertise | Information | Links | Privacy Policy | Support