Most people are familiar with basic reverb, delay and modulation effects, but what lies beyond? In the first part of a new series, Paul White explores the twilight zone of effects processing. This is the first article in a two‑part series.
It's often said that there's little to be had in the way of novel effects nowadays. Most effects are either standard reverb, delay, modulation or pitch‑shift, but that doesn't mean that there aren't other effects to be found lurking in the dark corners of your multi‑effects box or software plug‑ins folder. Some of the more bizarre effects have been around for years — for example, vocoders, ring modulators and chordal resonators — but the freedom provided to designers by the newer plug‑in formats means that more off‑the‑wall stuff is appearing all the time. The aim of this article is not to concentrate too much on specific products, but rather to explain some of these less common effects types and to make a few suggestions about the ways in which they can be used.
Ring Modulators are intriguing devices designed to process two input signals in such a way that the sums and differences of the input frequency components are generated while the input signals themselves are suppressed. For example, if you were to put in two sine tones at 500Hz and 600Hz, the output would comprise tones at 1100Hz and 100Hz. Conversely, feeding the same 500Hz tone into both inputs would produce components at 0Hz (a silent DC offset) and at 1000Hz (an octave up from the pitch at the inputs). However, the results are only as simple as this when you input pure tones — when harmonically rich sounds are used, all those harmonics contribute to the sum‑and‑difference process, resulting in a harmonically very complex output. Note that an output will only be produced from a ring modulator when signals are present at both inputs, so if level fluctuations are a problem then it may be worth compressing one or both input signals.
Because of the way in which the output frequencies are generated, ring modulators generally produce atonal, non‑musical sounds, which has made them popular for science‑fiction special effects, but they can also be used musically with a little care. For example, if two similarly pitched synth patches are ring modulated together then, providing the input waveforms are not too harmonically complex, the output can be both interesting and musically useful. Some non‑harmonic components will almost certainly still be present, and detuning the two inputs by a very small amount can produce unusual low‑frequency beating effects, but you can arrive at some very worthwhile sounds in this way. If you find ring modulation a little too strident for you, it can often be made more palatable by blending some of the original unprocessed input in with the processor's output. If you're still a bit cagey about using this effect, then perhaps the safest tactic is to use it as a sound design technique, sampling any isolated moments for later use — ring modulating a 100Hz tone with a vocal to produce the familiar Dalek voice is always fun, at least!
Processing percussion via a ring modulator can be good — use a pitched synth sound for the other input and you'll end up with a metallic, pitched drum part that could form the basis of an experimental electronic song or dance track. Ring modulating different cymbal sounds together is also an interesting experiment, which creates new, electronic‑sounding cymbals.
If you want to create new sounds and treatments based upon a basic ring‑modulation sound, try combining it with other effects. For example, use a dry sound as one input to the ring modulator and its reverb or delay as the other. You can also further process the ring modulator output using conventional but dramatic effects such as flanging or heavy delay.
At one time, vocoders were considered quite esoteric, but nowadays they come built into some multi‑effects units — less costly stand‑alone units are also fairly common. On top of that, there are some very effective vocoder plug‑ins that can be used within sequencers.
Like the ring modulator, a vocoder requires two inputs to generate an output, and to make this process clear, a block diagram is shown in Figure 1. Essentially, the vocoder superimposes the frequency spectrum of one sound (called the modulator) on a second sound (known as the carrier). The way this is achieved is that the frequency spectrum of the modulator is continually monitored using a bank of frequency‑spaced band‑pass filters, and the information used to control the gains of a corresponding bank of band‑pass filters in the carrier's signal path. Thus, as the spectrum of the modulator changes, the carrier's filter bank settings follow it. If a voice is used as the modulator and a harmonically rich musical sound as the carrier, this results in the classic vocoder sound — the voice seems to take on the pitch and timbre of the carrier sound, but the vocal articulation is still recognisable, because of the dynamic action of the filter bank following the continually changing spectrum of the voice. As you might imagine, the more filter bands the vocoder has, the more accurate and intelligible the speech‑like element of the output signal.
It is apparent from this description of the vocoding process that you need signals arriving at both inputs simultaneously before you can obtain an output signal. It can also be helpful to compress both inputs to the vocoder, in order to keep the output levels stable. On the other hand, if the modulation input is a voice, you might find that the vocoder is triggered undesirably by breath noises, in which case a gate inserted between the microphone and the vocoder's input will also be an improvement.
The talking synth effect has been used on countless records (for example, 'Blue Monday' by New Order, 'Mr Blue Sky' by ELO and 'Rocket' by Herbie Hancock), but this isn't the only way to use a vocoder. By substituting the vocal input with a recording of background noise in the local pub, and by vocoding this with a rich synth pad, you can create a very organic pad sound with a lot of movement. Similarly, two different synth sounds or samples can be vocoded together to create a totally new sound. If the modulator signal includes dramatic changes, such as a filter sweep, these will be imposed on the carrier. The real key is to experiment, but a point to keep in mind is that, because the end result is created subtractively, the carrier signal needs to be harmonically rich in order to give the filters something to work on. If the carrier is a synth sound, an open filter setting combined with a sawtooth or pulse wave works well.
When vocoders were first developed, it was realised that, while the pitched elements of vocal sounds provided a good modulation source, unpitched vocal components such as 'S', 'F' and 'T' sounds tended to get lost, and so vocal clarity was lost. Different strategies were devised to help with the intelligibility of vocoded speech. One of these was to replace some of the consonant sounds with bursts of noise, but a far simpler method was to add a high‑pass filtered version of the vocal input into the output. The latter method works well, because the higher‑frequency region of the vocal spectrum contains most of the energy of many vocal consonants, yet without many pitched components. By filtering out everything below 5kHz or so, then adding the remaining high frequencies to the vocoded signal, vocal intelligibility can be improved enormously. If this facility isn't already included in your vocoder, it can be patched up using hardware, or within a plug‑in environment that allows series and parallel routing (such as in TC's Spark).
Further sophistication is offered by some advanced vocoders where it is possible to swap around the filter bands, such that the level of the modulator input at one frequency can be mapped to control a filter band at a different frequency. Though few dedicated vocoders have this function and there's no simple way to fake it, implementations are possible using a software modular synth and it can really open up the gates of weirdness.
Almost everyone has tried setting a DDL to a very short delay time and then increasing the feedback. The result is a 'ringing' sound at a pitch determined by the delay time, particularly when it is excited by percussive sounds. For example, a 1mS delay will produce a resonant pitch with a fundamental of 1kHz, 10mS will produce a resonant pitch with a fundamental at 100Hz, and so on. With drums, it produces an effect not unlike playing them in a resonant brick tunnel — check your train timetable, though, before trying this at home!
A similar effect can be created by setting up a band of equalisation with high boost and resonance values — this will ring at whatever turnover frequency you select. If you have a number of bands available then you can set up a number of resonant peaks, which already holds a lot of scope for experimentation. However, if you can automate the frequencies of the resonant peaks, then you can get really creative.
Lexicon used the resonator principle in the PCM80 to create their resonant chord effect, and something similar was used by Alesis in their original Quadraverb. The concept was that MIDI note information from a keyboard or sequencer could be used to tune one or more resonators to specific musical notes so that any input signal passed through them would cause the filters to ring or resonate at musically relevant pitches.
Percussive sounds seem to work best with resonator programs and, because of the effect of the resonators and their musical pitches, the drum sound becomes more abstract and harmonic. Considering how dramatic this effect can be, it's surprising that it doesn't feature on more records, though I know that a number of sound designers use it for creating less conventional rhythm loops. If you have a multi‑effects unit with a MIDI controllable resonator, either monophonic or polyphonic, I'd recommend you try it out at least once so that you get to know the extent of its capabilities. It can help to make the effect more obvious if you increase the resonance or feedback, but otherwise I don't have many tips for experimenting with this effect other than 'suck it and see'!
I'll finish off for this month by looking at some of the triggered gating effects that can be set up. It's fairly well known that a gate can be triggered via its side chain to chop up audio in a rhythmic way, but it's sometimes possible to take this concept a little further. A MIDI gate or plug‑in is easiest to use for this purpose, as you can feed in a rhythmic sequence of note‑messages to trigger it. However, any regular gate with an external key input can be triggered using a fast‑attack, fast‑release synth tone.
Now let's look at some ways of making rhythmic chopping more interesting. One option I've heard used to great effect on vocals is to set up an even tempo‑related chop at around eight chops to the bar, then to use a DDL to delay a copy of the chopped signal so that the repeats fall exactly into the gaps created by the gating of the original part. Figure 2 should make it clear how this is done. What you hear is a kind of chattering effect as the repeated sections are joined up. The intelligibility is, of course, pretty poor, as half the signal has been discarded and the other half doubled up, but it makes for an interesting interlude. Pan the original gated signal and its delay to opposite sides for an even more dramatic effect.
If you're feeling adventurous, you could set up two or more gates triggered in such a way that each gate is only on for certain beats of a bar. Arrange your trigger material so that only one gate is open at a time, but with one of the gates open on each beat. This will require the generation of two or more MIDI note control tracks or, if the gate is being triggered from a synth, you'll need two or more different outputs to feed the gate key inputs. Finally, feed the same signal into all three gates, but then apply a different effect to each gate output. For example, use heavy flanging on the first output, distortion on another and perhaps an envelope‑following filter on the third. When the outputs are heard together, you'll hear all the differently effected beats spliced together. Figure 3 shows this technique using three gates. If the gates click during the transitions, lengthen the attack and release times slightly, but otherwise use the fastest settings for the cleanest chopping.
Taking the chopping up idea even further leads us into the murky terrain of granular synthesis and processing, where audio is sliced into extremely short pieces that are then joined back up in different orders to create new sounds. This requires special software, or ingenious use of a sampler —it's not something you can knock up using a multi‑effects box, certainly, and in many cases the process isn't even real‑time. I must admit that I've yet to hear anything musically worthwhile from processing of this kind, but if I discover anything, you'll be the first to hear about it!
That's probably enough weirdness for this month, but don't relax yet, as there's more to come next time when I'll be concentrating on filtering and 'time travel'.