To start off this short series on multi‑effects programming, Paul White explores the key building blocks found in today's versatile units. This is the first article in a five‑part series.
A little over a decade ago, effects units were mainly dedicated devices; you'd buy a reverb unit that produced only reverb, a digital delay line for delay or echo effects, a pitch‑shifter for transposition effects, and so on. This meant that we got to know the parameters relating to the various effects pretty well, but then along came the multi‑effects unit, offering all the above, and more, in a variety of configurations. With the multi‑effects unit came presets and easy edit options, which meant that a lot of newcomers to recording skipped over the essentials of effects creation. But creating a multi‑effects patch is a skill in its own right — you need to know how the different effects are created, and how they interact with each other, depending on their position in the signal chain. For the benefit of those who haven't studied the workings of effects before, I'm going to start by looking at the main effects types.
Reverberation is now a very familiar effect and has been discussed at length within these pages on numerous occasions, so I'll get through the essentials as quickly as possible. Natural reverb is created when sound reflects around a confined space such as a building, and the resulting reflection patterns are immensely complex. For example, a single handclap in a cathedral can generate thousands of reflections within the first second, and the complexity continues to increase while the reverb decays, as each reflection is re‑reflected from multiple surfaces. The first set of reflections from the walls and ceilings is known as Early Reflections; their pattern provides strong psycho‑acoustic clues to the nature of a space, even when you can't see it. In a large hall, the first few echoes may well be quite distinct before they build up into a dense reverberation, whereas in a smaller room the distances are much shorter, so the reverb density builds up far more quickly.
Though they're technically quite complicated, the low cost of DSP chips (Digital Signal Processors) has brought us a whole range of affordable reverb and multi‑effects devices. These invariably have stereo outputs which create a wide, spacious effect, even when the input signal is in mono.
The most obvious parameter of reverb is the time that it takes to die away (decay) after a percussive sound such as a handclap or drum beat. For contemporary music production, decay times of between one and three seconds are the most useful, though most units have a far wider range, to allow the creation of special effects. Reverb settings with strong early reflections and a fairly fast decay are an effective way of creating a wide stereo effect from a mono source, such as a voice or instrument, recorded with a single microphone.
In a large space, there is a delay between the original sound and the first reflections being heard — it takes around a tenth of a second for a sound to travel to a wall 50 feet away and then bounce back to the listener. Most reverb units have a pre‑delay parameter, and the effect of adjusting this is to vary the apparent size of the room without changing the overall decay time. It also helps give a mix more clarity, by leaving a breathing space between the original sound and the reverb that follows it. Reverb treatments are based on algorithms (in this case, a mathematical model simulating the reflective properties of a room), and a typical unit will include algorithms for halls, rooms, chambers and plates (a plate is a mechanical studio reverb device based on a large metal plate), as well as non‑natural reverbs, such as gated and reversed versions. Figure 1 shows how reverb develops from a percussive sound.
Soft furnishings in a room will absorb more high‑frequency energy, hence reducing the high‑frequency decay time, but a tiled room or stone cavern will reflect well at high frequencies, resulting in a very bright reverb. To emulate these environments, the basic algorithms include parameters for adjusting the relative HF and LF decay times. Brighter sounds work well with drums, electric guitars, acoustic guitars and pop vocals; more classical acoustic instruments often work better with a more natural‑sounding reverb.
Gated reverb was probably one of the first multi‑effect treatments, as it was first achieved by miking a drum kit in an extremely live room, then using a noise gate to cut the reverb off abruptly at the end of each beat. Examples of this treatment are to be found on many early Phil Collins tracks, though the effect is less fashionable in this context today. Most multi‑effects devices and reverb boxes provide gated reverb as a single algorithm, though some still allow you to create it from separate reverb and gate 'blocks'.
Reverse reverb is a burst of early reflections starting off at a low level and gradually increasing, before ending abruptly. This reverse envelope is responsible for the reversed illusion the effect creates. Figure 2 shows the envelopes of gated and reverse reverb.
Digital delay is the successor to the tape‑echo unit, a special type of recorder using one record head and between one and four replay heads. A continuous loop of tape was used on most machines to prevent the tape running out part‑way through a performance, and the principle was simply that any input signal was recorded, then played back a short time later via the replay heads. An erase head then cleaned the tape before returning it to the record head. The delay time was set by the tape speed and the head spacing; the inclusion of a variable speed control, and on/off switching for the replay heads, made various delay effects possible. Repeat echoes were achieved by feeding some of the output signal back to the input (as shown in Figure 3).
Digital delay, or DDL, is one of the key elements of a modern multi‑effects unit, and performs essentially the same task as the tape‑loop echo machine used to, but there are no tapes to wear out and the range of adjustment is far greater. It's also possible to modulate the delay time to create effects such as chorus, vibrato, flanging and phasing. Whereas most tape echo units were mono, today's digital effects are invariably stereo. A typical DDL effects block offers variable feedback, to produce multiple decaying repeats, and some offer multi‑tap delay, so several repeats at different delay times can be created. This is directly analogous to the multiple heads of the tape‑loop echo; feeding back some signal from all the taps makes the density of the repeats build up very quickly into a kind of pseudo reverb. Figure 4 shows how a multi‑tap delay operates.
Further interest can be created by arranging for the repeats to come from alternating sides, or by feeding some of the left‑channel output back into the right input and vice versa. These options are usually presented in the form of 'algorithms', where the routing is set, but the user generally has control over the values of various parameters, including delay time and feedback. There's also a mix parameter or control on most multi‑effects units, which allows you to adjust the balance between the original sound and the delayed sound.
The simplest DDL effect is a single repeat using no modulation and no feedback; the delay time can be varied up to the maximum range of the unit. Short delays of between 30 and 100ms are used to create slap‑back echo effects or pre‑delays for reverb effects, while longer delays produce a distinct single echo. Echoes timed to coincide with the tempo of the song can sometimes be effective.
You can get multiple, equally spaced, delays by increasing the feedback value. At longer delay times, you'll hear the familiar repeat echo effect but as the delay time is shortened, you'll notice that the echo effect disappears and is replaced by a metallic resonance, sometimes called tunnel echo. The frequency of resonance depends on the delay time; increasing the feedback will maximise this resonant effect. Several good examples can be found on the more adventurous dance records and dance‑loop sample CDs.
Using a multi‑tapped delay lets you create a less rhythmic echo; for the best effect, the delay times of the various taps should not be set to exact multiples of each other. Adding feedback causes the echo decay to become quite complex, and this effect is popular on electric guitar (from the Shadows to U2), on vocals, and on instruments such as synth lead lines, sax and flute. As we'll see later in this feature, combining delay with reverb often produces the most interesting results.
Modulating the delay time using an LFO makes the pitch of a delayed signal waver both sharp and flat at the rate set by the LFO speed. The depth of modulation determines how far sharp or flat the sound goes. The simplest modulation effect is pitch vibrato, where only the processed (sometimes called 'wet') sound is used — the original, unprocessed ('dry') sound is turned off using the mix parameter. The sound you hear from the output will be delayed slightly, but if the delay time is set to less than 10ms, it will be too short to notice. If you set a modulation rate of, say, 4Hz, then turn the depth control up slowly, you should hear the sound being processed take on a wavering effect, not unlike that produced by the mod wheel on a synth.
To convert vibrato to phasing, set the mix parameter to give equal amounts of dry and delayed sound and experiment with delay times between 1 and 10ms. As you adjust the modulation speed and depth, you'll hear the individual harmonics that make up your sound moving in and out of phase with each other, which has the effect of filtering the sound in a very dynamic and complex way. This is known as comb filtering, because a frequency graph would show lots of narrow spikes and troughs, rather like the teeth of a comb. As the delay is modulated, these 'teeth' move up and down the audio spectrum, and you'll probably recognise the effect as being similar to what you get from guitar phaser pedals. You can vary which harmonics are affected by changing the basic delay time; the shorter the delay time, the higher the frequencies that are affected, and vice versa. If your unit has a feedback invert function, try switching this in, as it too affects the harmonic structure of the effect.
In the context of a multi‑effects unit, modulation effects are likely to have their own algorithms, rather than relying on the user to create the effect from scratch using a general DDL block. Irrelevant parameters will be normally be excluded and the delay‑time range restricted to that relevant to the particular modulation effect being created. This helps steer the non‑experienced user in the right direction, and may allow some other non‑standard parameters to be added by the manufacturer.
Flanging was first created by simultaneously running two tape machines carrying copies of the same music, then mixing the two outputs. If the two machines are perfectly in time (sync), with each other the two signals add normally, but if the timing between the machines drifts for any reason, you hear a phasing effect due to comb filtering. By deliberately slowing down one machine and then the other, using hand friction on the supply reel, you can control the phasing effect, though it takes some practice to get it right.
Flanging can be simulated digitally, though the result is never exactly the same as you get from doing it the hard way. DDL flanging doesn't exactly duplicate tape flanging because the delay can never pass through the zero‑delay point, as it does when one tape machine overtakes the other. It's possible to fake this so‑called 'through‑zero flanging', though, by using two DDLs. This is achieved by setting a fixed short delay on one DDL, then arranging for the second DDL's delay time to be modulated either side of this value. The settings for flanging are similar to those used for phasing, except that longer delay times — typically 10 to 50ms — are used, and the feedback value increased to make the effect more resonant. On more sophisticated units, it may be possible to modulate the delay using some source other than the LFO — for example, the input signal's envelope. This can help make the effect less obviously repetitive, and hence more like its tape counterpart. Extensively used in the psychedelic '60s, flanging is making a bit of a comeback in dance music, but if overdone it can become a cliché, so it's often best combined with other effects, as will be described later.
Inverting the phase of the signal fed back to the input allows different harmonics to be accentuated by the filtering process, just as it does with phasing. Try both options and see which you prefer.
The chorus effect is essentially the same as vibrato, but with an equal proportion of the dry sound mixed in. The idea is to produce the illusion of two instruments playing together, where the ensemble effect is created by slight timing and pitch differences. By setting a longer delay time than for vibrato, between 30 and 150ms, say, you can make the effect of the timing differences between instruments more pronounced. You can also use very gentle modulation on even longer delays to create a combined chorus and echo effect. The modulation speed is usually set in the range 2‑6Hz and the depth set by ear so as not to sound too out of tune. The longer the delay time, the less depth will be needed. More sophisticated chorus effects blocks may include multi‑tapped algorithms, which produce the effect of multiple chorus devices operating at the same time.
Chorus was first used on electric guitars and synth string machines, but it can be used on virtually anything, from fretless bass to synth pads. However, it doesn't usually work too well on vocals: the effect is too regular to sound entirely natural. Though most chorus effects now come as stereo, you can fatten up a mono chorus by panning the original, untreated sound to one side and the modulated delay to the other. The result is a moving wide sound source that seems to hang between the speakers.
Many multi‑effects units include an algorithm to simulate rotary speaker cabinets, devices originally used by organ players to add an interesting chorus/vibrato to their sound. The cabinets work by means of a rotating baffle around the bass speaker, and a rotating horn to carry the high end. A motor‑and‑pulley system provides two operating speeds; because of mechanical inertia, the system takes a finite time to switch between these two speeds. In a multi‑effects unit, the effect is usually created by a combination of modulated delay and filtering but, as far as the user is concerned, there may only be 'fast', 'slow' and 'off' speed options to select from. The inertia effect is simulated by using a special LFO that changes speed over a period of a second or so rather than instantly, and the skill of using a rotating speaker (or emulation) is to operate the speed change at appropriate points during the performance to provide the right feel. This effect also works well on guitar, synth pads, and sometimes even on voice.
A panner is simply a device that automatically pans a mono signal from left to right and back again, usually under the control of an LFO. It is closely related to tremolo, an effect created by modulating the level of a signal using an LFO, except that in the case of the panner there are two modulating circuits, one of which is at maximum gain while the other is at minimum — as shown in Figure 5. Some effects units incorporate options for triggering the pan to MIDI notes or sync'ing it to tempo via MIDI Clock, in which case it's possible to synchronise the panning to the tempo of the music. While panning raw sounds around the place is a little dated, it can be effective to leave the original sound where it is and just pan the reverb or delay element of the sound.
A more recent advance is three‑dimensional panning, an effect that combines left/right panning with psycho‑acoustic front/back panning. Continually changing levels and EQ changes make the signal sound as though it's moving in a circle in front of the listener rather than simply following a straight line. As the sound pans, it seems alternately closer to the listener, then further away.
Pitch‑shifting is a process for changing the pitch of a sound without changing its length. (If you change the pitch of a recorded sound by speeding up the tape, the shifted sound will, obviously, get shorter.) The range is usually one or two octaves up or down, with fine tuning as well as semitone step adjustment.
The effect works rather like an automatic sampler that samples very short sections of the sound, loops these sections, then plays them back at a different speed. Clever algorithms are used to hide the splicing between subsequent loops, but, on all but the most sophisticated units, the splicing shows up as a warble when large amounts of pitch‑shift are used.
Most multi‑effects devices include a pitch‑shifting section, which may allow simultaneous shifting up and down in pitch. If small shifts of between five and ten cents, both positive and negative, are applied to a signal, you get a nice alternative to chorusing without any obvious cyclic sweeping. It's also common to find an option for adding delay to shifted sounds, and for feeding the shifted output back to the input so that it will be shifted again. This can be useful as a special effect to create fixed‑step arpeggios.
Such a basic pitch‑shifter is of little use in creating true musical harmonies, because the only useful fixed intervals tend to be octaves and fifths. However, some models now include intelligent pitch‑shifting, where the user defines a musical scale, and the device automatically picks the correct harmonic interval. This is done by tracking the pitch of the input signal (which must be monophonic for clean tracking), then applying the correct amount of shift to produce the desired harmony note. The musical scales may either be presets or user‑programmable.
Though 'formant‑corrected' pitch‑shifting is not yet available in typical multi‑effects units, it can only be a matter of time before it becomes standard issue. With a conventional pitch‑shifter, moving the pitch up creates a Mickey Mouse effect, because not only does the musical note change, but the timbre of the sound also changes, just as though you were speeding up a tape. But when a real singer sings a new note, certain components of the sound actually stay the same — their throat and chest cavity resonances remain constant. Using formant‑corrected pitch‑shifting, it should be possible to move notes to any pitch but leave them sounding natural. Alternatively, you could deliberately move the formants to make a male voice sound female or vice versa.
This is going to be a hugely important area in the not‑too‑distant future, and when something affordable arrives, we'll be the first to let you know! At the moment, the only affordable formant‑corrected pitch‑shifters are built into hard disk recording systems and require off‑line processing, though Digitech's Vocalist range of devices, designed by IVL, employs a couple of clever techniques to help make the shifted sound seem more natural.
Next month I'll be looking at the more common dynamics and EQ‑based processing blocks before focusing on how the different blocks work together.
A vocoder is a special type of multi‑band filter that automatically mimics the frequency‑spectrum characteristics of a signal being fed into its control input. You can visualise it as a graphic EQ that's able to listen to any signal, and constantly adjusts its fader settings to match the spectral content of that signal. When a signal is passed through the vocoder's filters, it takes on the same spectral characteristics as the control input, which is how the effect of talking keyboards is produced — but there are far less clichéd ways of using vocoders. For example, you can trigger them from more rhythmic sounds to turn a pad synth into a melodic rhythm. Some of the better sample CDs make extensive used of vocoders in unusual and interesting ways, so check out a few of these for inspiration.
Pioneered by Lexicon, the resonant‑chord effect is achieved by using several delay lines, each set to a very short delay with feedback, and each tuned to resonate at one note of a musical chord. Any harmonically rich sound fed into these resonant delays will 'ring' at the appropriate notes, and if you use MIDI control to change the pitch at which the delays resonate, the ringing‑chord effect can be played from a keyboard. The effect has to be heard to be fully appreciated; depending on the input source, the results can range from ethereal to disturbingly mechanistic.