The image above looks totally awesome, and it’s a picture of a tape delay plug-in, which is a time domain effect, as are modulation effects. However, tape delay is not a modulation effect since there is no modulation occurring. Unless you have more than one tape machine…
OK, so wow and flutter are modulating the sound somewhat (imperfections related to fluctuating tape playback speed and pitch) but tape delay is not a modulation effect.
Today we’ll move on from the delays and check out some of delay’s time-based effect relatives, the modulation effects like chorus, phase, vibrato, tremolo, auto-pan and good old flanging.
We’ll finish with the topic of time-based processing by discussing reverb types and parameters and uses tomorrow and the next day. For now, we’ll discuss the modulation effects.
Apart from reverb and delay, the other main class of time-based effects are these modulation effects. They all rely on an LFO to modulate the sound in a rhythmic way, which is why they are classed as modulation effects.
These are basically effects where you clone a signal and then move that copy (or, more often, multiple copies) in time relative to the original. You mix the copies together with the original to thicken up the sound or add a sense of movement and/or width to the sound. An example would be a wide, shimmering chorus effect on a vocal or an electric guitar part.
Some of these effects are simply variants of delay on the order of a few milliseconds, where the delays are subtly tuned away from the original pitch, up and/or down by maybe a few cents.
Others involve a different sort of cyclic behaviour, for example, auto-pan or tremolo effects. There are yet more types sweep through frequencies to add a feeling of motion by inducing continuous tonal changes.
These effects are not used on drums very often, because the nature of the effect requires the sound to sustain rather longer than drum and percussion sounds typically do.
The sound to be processed should have a decent length to its sustain phase (as in ADSR envelope, see my synth blogs) for the processing to be worthwhile.
In short, your best bet for a sound that works well with modulation effects is one that sustains for at least a handful of milliseconds, maybe 50ms and ideally much more.
For truly lush effects, the sound you want to process will ideally sustain for at the very least half a second most of the time it is present, allowing the LFO time to do it’s thing for a decent stretch of time and provide a noticeable musical effect.
ADT (Automatic Double Tracking)
If you record two takes of a performance by the same person and they are intended to be identical, there will still be subtleties that are unique to each performance. A take cannot align perfectly with another in every respect, no matter how gifted and diligent the performer is.
The reason it sounds good to double-track something closely is actually those imperfections. Identical takes would not be helpful, but you do need to be pretty close to matching the original to not sound messy, especially if you want the effect to be subtle.
The Beatles made the first use of “automated double-tracking” when engineer Ken Scott at Abbey Road Studios, in London, England in the mid-Sixties, invented a box to do just that.
He did this to satisfy John Lennon, who hated the process of trying to double a take and also preferred not to work too long on a recording in order to keep it fresh.
The primitive ADT effect matured over the next decade as people experimented with weird and wonderful ways to mess with audio by duplicating it and altering the copies subtly. This was very much in the spirit of the times.
This experimental approach to effects created a whole family of widening, shimmering, thickening and doubling effects that all involve modulation of the sound. Perhaps the most popular of these is the chorus effect.
A chorus effect uses delayed copies of the signal to create it’s effects. Delay times range from 10ms to around 30ms.
Using one copy, or even a whole cluster of quiet copies, can support a sound really well given there are differences between the copies as well as with the original.
This effect can be obvious and wobbly as in many recordings by the Police from the 1980’s, if the pitches are far enough apart, or it can be virtually transparent. Chorus can work to support a sound very well, but it’s usually most effective when it’s not too obvious. It can make people feel a bit seasick when it’s overdone.
Chorus makes copies of a sound and delays them between 10ms and 30ms, and then changes the delay time and the amplitude (level) over time to maintain a sense of motion and difference.
There are three controls that you’ll find with chorus plug-ins or pedals or rack outboard effects. There will be a waveform selection of some sort to set the LFO modulation waveform, and there will be a frequency control (modulation rate) and a depth control (sometimes called delay).
Of course, there will be level controls and a wet/dry mix control. There may even be multiple voices which can have these parameters set for each voice. Chorus plug-ins can get quite fancy.
It won’t really sound like a bunch of people playing the same thing together, as some aspects of the sound are too fixed, for example modulation rate and LFO waveform. It’s a great effect though, and very handy at times on guitars and background vocals to widen and thicken them at mixdown
Typically I would record this effect with a guitar pedal on a guitar, but use it in mixing as a plug-in after the fact on things like vocals.
Way back in the 1950’s, flanging was an effect you could get by applying gentle pressure to the flange of a tape machine while a second machine was running alongside with the same sound playing on both machines.
The subtle differences in the timing of playback between the two machines, never able to be properly synchronized, rendered a thin, hollow, sweeping effect that was later dubbed flange in the 1960’s at Abbey Road, again involving the Beatles and their technical team.
I have seen Les Paul credited by some with the first use for deliberate effect of the flanging sound, and this seems reasonable as he was an inventor and guitarist and artist, and he was the originator of many of the techniques of multitrack recording we are still using today. Flanging is probably one of them.
Flanging is a time-domain effect that relies upon the comb-filtering effect.
Comb-filtering is a hollow, thin sounding effect that occurs when two signals are mixed together with one of them delayed by a very short amount of time, just a few milliseconds in fact.
Due to the laws of physics regarding the behaviour of wavelengths and frequencies, certain frequencies within the combined signals will suffer from phase cancellation, getting far quieter, and certain adjacent frequencies will benefit from a summing of energy, thus reinforcing those frequencies.
There is a type of EQ circuit or plug-in called a notch filter, which exists to remove a user-selected frequency from a signal as cleanly and narrowly as possible, in a very targeted manner. This allows a notch to be made in the EQ spectrum of the sound.
It is intended to affect neighbouring frequencies as little as possible. Technically, it’s a very narrow filter with a very high Q value. Clearly, this is a surgical EQ, a tool meant primarily for corrective equalization.
Picture a series of notches in an an otherwise normal sound’s frequency spectrum. These notches are deep and narrow and evenly spaced, looking much like a fine-toothed comb.
This is why it is called comb-filtering.
Regular readers may remember that I discussed the relationship between octaves and frequencies earlier this month.
To quickly recap, every note will sound an octave higher if it’s frequency is doubled.
This relationship exists throughout the musical spectrum. At a frequency of 110Hz, the note ‘A’ sounds. The next ‘A’ an octave higher lives at 220Hz, and the next A after that will be at 440Hz. The next is, of course, at 880Hz. And so on.
This is not the case with comb filtering. In the comb pattern, the notches still maintain a mathematical relationship but it is simply identical sized spaces rather than a doubling of frequency. Using the same fundamental frequency example of 55Hz, the comb-filtering series would go from 55Hz to 110 Hz 165Hz to 220 Hz to to 285Hz to 330Hz and so on, keeping exactly the same 55Hz difference in the spacing between each notched frequency area.
Flanging and phasing both rely on this comb-filtering effect. It is actually the delay time for the delayed signal that is being modulated, and this makes the comb-filtering effect gain a sense of motion. The delay time changes caused by modulating it mean that the notches shift around too, because the phase relationships between frequencies is changing with the time delay changes. The idea is to make the variations in phase change in a cyclic pattern, due to cyclic modulation of the delay time by an LFO, and this rhythmic pattern of changing comb filtering patterns provides the classic flanging sound.
For flanging, use delay times between 1ms and 10ms. Keep the speed or rate of the LFO relatively low for the most useful effects.
Parameters will include a control to select a delay time, and rate and depth controls for the LFO that will modulate that delay time. There will also be the usual level controls, a feedback control, and a wet/dry mix control.
You can also set a frequency range for the sweep of the LFO modulation on many flangers, which helps a lot when you want a swept midrange sound and don’t really want to weaken the low end of the sound with sweeping flanger effects.
The waveform for the modulation is usually selectable too.
This means you will have a way to confine the range of the sweep to a part of the midrange rather than having it act across the entire sound, top to bottom.
Often there are two delayed channels rather than one, each modulated by the same LFO but with the LFO beginning at a different point in it’s cycle for each delay channel. This gives an even wider feeling to the sound. You may have multiple delays created, and the time delay is key to the effect.
Modulation effects tend to be used as insert effects (returning to and from channel insert points) or as in-line guitar pedals rather than as auxiliary send effects (buss-driven), but they work fine both ways. Obviously, you will want to use the 100% Wet setting if you are using chorus effects via an auxiliary send or buss.
Time delay is flanging’s secret weapon, but it’s close cousin phasing uses a very different method to generate it’s effects.
Phasing works by notching frequencies in octave relationships rather than the numerically even spaces between notches found in the flanging effect.
Thus, for example, 110, 220, 440, 880 would be a valid series of frequency notches in phasing, rather than the 110, 220, 330, 440 numerically evenly-spaced series found in flanging.
It uses an all-pass filter to generate the notches. This lets all frequencies pass, as you would expect from the name, but it changes the phase of some of them before recombining with the original sound.
The LFO modulation is applied to vary the notches, as with flanging, making phasing and flanging closely-related modulation effects. The LFO waveform used is usually a sine wave in phasers, so you don’t often get the option to change that as you would on most flangers.
Phasers have stages which are typically found in pairs. It takes two of these stages to make an inverted phase all-pass filtered signal and combine it with the original, and controlling the tonal character of the phaser is largely done by selecting the number of stages to be applied. There may be four, six or even eight stages in phasers.
You will find the same sets of controls on phasers and flangers in large part, although the delay time parameter on a flanger is replaced on a phaser by a control for selecting how many stages you want. There will usually be a width control if it is a stereo phaser – even guitar pedal phasers usually have two outputs for stereo use.
I think I can sum these up with the expression “chorus on steroids”.
With ensemble effects, you get a whole bunch of chorus effects – multiple voices all slightly differentiated by pitch and timing. These are all modulated by an LFO, as in chorus. However, the sound would just be like chorus if all the delayed voices were being modulated by the same wave shape.
In order to make a bigger, more convincing sound, the LFO starts it’s cycle in a different location in the waveform for each voice it modulates.
This means they all modulate at the same rate, but the phase relationships are rotated by so many degrees of phase amongst the various delayed voices, such that one voice may have a 180 degree relationship relative to another voice, and a 90 degree relationship to another. This is in terms of relative phase, one wave to another.
Due to the complexity of the cyclic behaviour, the character of an ensemble effect is wide and shimmering. It is not entirely realistic, of course, because the subtle differences between voices in the real world would be far more complex.
You’ll get great results by using ensemble effects in subtle amounts, as they are a bit over the top otherwise!
VIBRATO AND TREMOLO
Even though these are effects that have been around for many decades, there’s still confusion about what vibrato is, as opposed to tremolo.
Vibrato is pitch modulation, where the pitch goes up and down repeatedly in rapid see-saw fashion.
Tremolo is amplitude modulation, where the volume of the sound goes up and down repeatedly, in rapid see-saw fashion.
It didn’t help that there was a huge boom in guitar sales from the era of instrumental hits like Telstar and Apache by the Shadows in 1960 that saw the rise of the musical instrument retailer of today. The Shadow’s lead guitarist Hank Marvin was everywhere using his whammy bar with taste and finesse to decorate his melodic phrases.
Sadly, the whammy bar fitted to these suddenly addictive electric guitars was called a Tremolo Arm in those days. Kids called them tremolo arms rather than whammy bars, a more modern expression. Well, whammy bars modulate pitch, not volume. That makes them vibrato arms rather than tremolo arms.
Ah well. That’s guitar players for you!
Tremolo plug-ins appear in all DAWs I’ve ever used (and typically these are included when you buy the DAW software) but vibrato seems rather harder to find in plug-in form.
Fortunately, vibrato effects are easy to implement (and ubiquitous) on a synth or synth plug-in. Using an LFO set to modulate pitch generates vibrato as the LFO modulates the oscillator’s pitch up and down at the specified rate.
It’s helpful, of course, that you can also set the LFO mod rate to generate vibrato that is in tempo with the host DAW’s session tempo.
Personally, I like vibrato effects in shimmering 16th or 32nd rhythms, especially if they slow down slightly over the course of a bar, so they begin by seeming to be in tempo, but then draw attention and add some attitude by dragging a little over the rest of the bar. Cool.
As usual, there’s an LFO behind the scenes modulating away.
By default, it’s usually set to a square wave, which suits the effect when it’s blazing away full-on. There might be no other option.
Often, though, you will be given a choice of waveform, possibly called Shape instead. This will let you set up a gentler effect altogether, by using sawtooth, sine or triangle wave instead, since square waves are basically abrupt in nature, being a sort of cycling on/off switch. There could be a phase offset control if it’s a stereo unit.
Auto-panners are modulation effects too, and they also use the rate and depth approach using an LFO.
Using the phase control on a tremolo plug-in, you can invert the phase of one side of the stereo relative to the other, making it 180 degrees out of phase. This causes the sound to move from side to side of the image with the effect depth and rate, so you can use it for an auto-panning effect, rhythmically panning between the left and right channels of the stereo image.
Choose the smoothest waveform available, and set your rate control to a musical setting. Set the phase to be 180 degrees apart between the two channels. To get the hard-panned effect, set depth to 100%. Now back it off a bit, say to 90%, so it will always be present to some degree in both channels. This will also make it less disturbing for the listeners who use headphones. Auto-panning can be an irritant to headphone users, as you can well imagine.
There are dedicated auto-panners and they feature all the controls you would expect from the above description, but depth may be called spread or width on some of these auto-panners.
ROTARY SPEAKER/LESLIE SIMULATORS
For over fifty years, the Leslie sound has been a big part of the magic on innumerable recordings.
It is a sound generated from a large speaker cabinet containing a rotary speaker system invented by Donald Leslie. It usually gets paired with the Hammond B3 organ and appears on many a late 1960’s Beatles record, usually used to process a vocal or electric guitar.
It will give you back pain in the blink of an eye, which is a great reason to model the beast in software.
The Leslie is indeed the bane of roadies and B3 players everywhere, being amazingly heavy to move, but it sounds utterly gorgeous, so they have to carry it onstage anyway. This beefy characteristic has made the plug-in versions extremely popular!
The classic Leslie speaker cabinet actually contains two speakers. There’s a horn tweeter at the top of the cabinet, and a bass/midrange speaker at the bottom.
The horn at the top is actually a pair of horns mounted opposite each other that spin around under motor control, but only one emits sound. The other simply acts as a counterweight during rotation.
The bass/midrange speaker in the lower part of the cabinet does not rotate, due to the inertia that would have to be overcome to get it’s large mass moving, and a similar issue arises when you try to stop the thing rotating.
The mass of the magnet on the back of the lower speaker is too great to effectively rotate the speaker assembly.
The solution Donald Leslie came up with was simple and ingenious, like all great ideas.
He rotated a drum around the stationary speaker instead, leaving a vent in the drum for the sound to get out through whenever the vent passed in front of the speaker.
As the drum rotates around the lower speaker, the low and midrange frequencies all get louder whenever the vent is passing in front of you, and then get quieter as the sound is blocked off by the drum. This repeated action causes a tremolo effect (amplitude modulation) due to the differences in volume emitted towards you from that speaker as the drum rotates.
There’s also a pitch modulation due to the Doppler effect, which I’m sure you’re aware of, whereby the sound from the speakers approaches, passes, then goes away from you and so on, in a cyclic manner. We’ve all heard the sound of an approaching ambulance drop in pitch as it passes by at speed. That’s the Doppler effect.
The amplitude/pitch modulations causes tremolo and vibrato effects, and distance perception (back/front and side/side) is altered cyclically too. This back/front modulation is strongest in front of a real Leslie at the same moments that the pitch modulation from the Doppler effect is at its strongest. This makes for an interesting, complicated and very musical sound that will stand out without having to be aggressively balanced in the mix.
The system is a two-speaker system, os emulations have to split incoming audio into two discrete frequency ranges, one for the high mids and highs, and one for all other frequencies below the crossover point. This is the frequency at which the two systems centre their overlapped area of frequency response. It’s never an exact point, more a narrow range centred on a specific frequency where both speakers will emit those frequencies. It’s a gentle but narrow window in the handling of the crossover frequency area between the two speakers, ideally, and as transparent and artifact-free as possible.
All of this makes the Leslie a complex effect to emulate in software, since it involves aspects of tremolo, vibrato, chorus, phase rotation, all of which also involves LFOs, synthesis and auto-panning. It sounds amazing though!
There will usually be a room sound simulation control of some sort to add reverb. In the real world, a real Leslie cabinet would be throwing sound waves around and bouncing them off nearby surfaces. This generates an intensely complicate reverberant field with many common materials in rooms, such as wooden floors and flat ceilings. Room sound is a part of the Leslie sound.
Players can also adjust the speed of their Leslie’s rotation using a slow/fast lever, that controls the acceleration and deceleration of the rotors in the cabinet. This lever appears onscreen in all the Leslie emulations I’ve seen, and is essential tot he playing style of many great organists. It takes time for this effect to occur, due to the physics involved, and that small delay in the Leslie’s response time when adjusting the rotor speed is emulated in the software too.
The sound seems to come from left to right, then from right back to left when you listen to a Leslie in action. That means you will find auto-panning functionality in your Leslie emulation.
Thanks, Donald Leslie.
Exterminate! Exterminate! Destroy the Doctor! Exterminate!
The Daleks spring to mind the instant that ring modulation is mentioned. It’s such a cool sound!
It’s a hollow, metallic, thin sounding effect, and therefore useful for somewhat inhuman or disturbing sounds. It was one of the earliest synthesis techniques to be developed, dating back to certainly the 1950’s.
All it involves is modulating the level (amplitude) of one waveform with another. The output produces two frequencies, being a sum frequency and a difference frequency, but does not output the original two frequencies. If you modulated a 300Hz waveform with a 75Hz waveform , the ring mod’s output would be the sum, 375Hz, and the difference, 225Hz.
You won’t be wanting this sound very much. It’s only so useful, but it is pretty cool to fool around with.
It’s not the most controllable effect in the world. There will be a frequency control to set the frequency of the oscillator, if it has a built-in oscillator. Some means of selecting an alternative modulation source is common in modern plug-ins, possibly with a carrier wave and modulator wave setup, but using the sidechain feature of the plug-in, if present, with a spare DAW buss (mono or stereo) is the simplest and, by far, the most flexible way to do this.
Have fun with it. It may be a while before you think of using it again!
These do what it says on the can. You put in a waveofrm, and it shifts the frequencies present in the signal by a fixed mathematical amount up or down.
Looking back to the octave to frequency relationship discussed recently in this blog series, you might recall that octaves have a doubling relationship (110Hz, 220Hz, 440Hz and 880Hz are an ‘A’ note ascending an octave at a time in pitch).
Also recall that comb-filtering had fixed numerical spacing generating the “gaps” between the regularly spaced “teeth” of the so-called comb filter.
For example, 110Hz, 220Hz, 330Hz, 440Hz, 550Hz, and so on. Increasing in frequency 110 Hz at a time in this particular example, rather than in a doubling ratio as in the natural harmonic series of sounds heard in nature.
Comb-filtering is inharmonic by definition, and it makes things sound thin, hollow and metallic, or “phasey” if you will.
It’s the result of the sum and difference aspects of the signal, the alternating regular amplitude summing and cancelling being spaced like steps in a ladder up the spectrum. Teeth of a comb. You will recall the phaser effect employed comb-filtering in stages to generate this unnatural inharmonic series. The result was a series of “comb teeth”, or a ladder, if you like, that climbed the frequency spectrum.
This is the same thing a frequency shifter does. You pick a shift value, say 100Hz, and the output is a series of adjustments to the overtones heard in the wave based on the shift value. These are good tools for sound design, especially for alien sounds. They can produce sounds quite like ring modulation, very resonant and metallic and hollow.
That’s it for time-based effects, you might think. Actually, the next two blogs will be on the topic of reverb, the most complex and ubiquitous time-based effect of all. See you there!