Here are 15 delay techniques from simple and subtle to straight up whack.
1. Faux Reverb
I hinted at this in my post about using reverb to make a violin pickup sound more acoustic. In short, this is what I use as an alternative to a reverb effect.
The idea is that we are far less likely to notice sounds that have the higher treble frequencies filtered out, so using a low pass filter (sometimes shown as ‘high cut’) on a delay makes the effect far more subtle. A stereo delay with the highs rolled off is also a popular trick used by mixing engineers to add depth to a recording – often a vocal take – without sounding like it’s in the same space as the rest of the tracks.
I’ll generally use a stereo ‘ping pong’ delay that has different lengths for the left and right channel, at 500ms or less. The filter can be lowered down to 1khz to make it blend nicely.
2. Stereo Double
Double tracking is a useful technique that uses two recordings of the same sound and spreads them out into the left and right speakers. The idea here is to do something similar, but live – use a short delay on one channel in a stereo setup to give it the sound of a second copy of you playing slightly behind you.
We’re talking about 10 or 20 milliseconds here, so that the delay isn’t noticeable as its own sound but more as an extra copy.
This should only be used in situations where you have complete control of the listening environment, as mono speakers such as a phone or Bluetooth speaker will sum the two copies together and introduce comb filtering. More on that in the ‘Binaural Shenanigans’ section further down.
3. Long Dark (Palettes Intro)
While the faux reverb is best with shorter time, this is where we can start to really push the length and still enjoy a subdued effect.
In the first track of Palettes (my album with Reuben Leng), there is a very subtle, almost choral pad that can be heard in the background. Here is a brief version of that track:
It’s not a synthesized part, but rather a long 900ms delay (in series with a 500ms, 600ms and 800ms delay panned separately) from the violin track with the high frequencies filtered out and the feedback turned up to 99%.
4. Compressed Repeats
When you want long sustained tails from the notes you play, the first solution is to increase the delay feedback. This works great, until you introduce some new notes that are dissonant with the delay trail, ending up with a heap of spicy notes that just won’t go away.
A neat alternative is to lower the feedback to, say, 80% and use a compressor or limiter after the delay to keep the trails sustaining. When you play new notes, the old ones seem to fade away quickly and let the new notes take over.
This is, once again, a small section from the same album. This is heard in the final track called Daisy Cutter:
5. Loop Delay (8 seconds is 4 bars)
There are only a handful of delay pedals that are capable of long delay times, and to be honest they don’t really need to be long in most cases.
When you do have a delay with upwards of 8 second delays, it essentially becomes a looper and does looper things.
One of my favourite videos as a young violinist was seeing Ed Alleyne-Johnson busking in Chester, England. His delay pedal in that video is set to an 8 second delay, meaning that it is exactly 4 bars long at 120BPM. This way, he can turn on the pedal to add a new layer, turn it off to let the repeats continue, and lower the feedback to have the loops gradually fade away over time.
Not many loopers have the ability to adjust the feedback, so this can be a great alternative (assuming you don’t want to save your loops later).
6. Rhythmic Delay
I often use a rhythmic delay when I’m playing live and need to stay at a particular tempo for my piece. This way I don’t need to rely on having a click track, although it does mean I have to pay more attention to timing my notes evenly.
One of the more common rhythms used by guitarists is the dotted 8th note delay. When playing 8th notes, the delayed signal is displaced so that it now subdivides the 8th notes. The image below can probably explain it better.
7. Polyrhythmic Delay
Why stop at just one rhythmic delay, when you can have many! I’ll often have multiple delays with their own (read: parallel) signal path, but having them feed into one another can also start a rhythmic chain reaction. Having the different delays coming from different speakers is great fun, and it’s where surround sound really starts to get addictive.
I have a section below that talks about how to calculate the delay time for different subdivisions, however the short and sweet alternative is to multiply the delay time by the ratio you want to achieve.
For a 3:4 ratio – a 3 against 4 polyrhythm – my delay time might look like this:
500ms x (3/4) = 667ms OR 500ms x (4/3) = 375ms
Both will give you a 3 against 4 rhythm.
8. Faux Tape
Some delay algorithms emulate the action of an analogue delay pedal, where changing the delay time causes the repeats to rise and fall in pitch as they are sped up or slowed down. Being able to modulate the delay time with an LFO gives it an over-the-top wow and flutter like a warped LP or bargain basement cassette deck.
I’m not really after a convincing tape echo emulation here, but more of a lo-fi delay. Having the delay signal go through saturation and filters can give it some very colourful sounds.
A Bit Odd (The Fun Part)
9. Glitch Delay
This takes the idea of the lo-fi delay, and applies it to a digital-style algorithm instead. Rather than the pitch rising and falling, changing the delay time causes the signal to jump forwards and backwards in a stuttering mess.
I have a patch that keeps the delay time the same while I’m playing, then increases the delay time when the signal is quiet.
10. Stereo Reverse
This is usually something that you need to cook up yourself, as most reverse delay effects in the wild are strictly mono.
In this case, having the ability to have the reversed signal only in one side of the stereo field gives lots of movement across channels and gives the listener something to follow after the initial sound has stopped.
11. Anti-Delay (the repeat lands before the next note)
The neat part about rhythmic delays from before is that they can subdivide the rhythm you play and make an interesting cross rhythm.
If the delay has the same timing as the rhythm, then they land at the same time in rhythmic unison. If the delay is slightly short, then it lands just before the next note – it now appears to come before the notes are played.
If the rhythm is consistent, then the anti-delays keep building up and take on an almost reverse-like character where the repeats ramp up in volume before the note is played.
12. Harmonised Delay (4 Part Harmony Patch)
This is achievable in two ways: (1) with two stereo harmonisers, or (2) with a modular synth where each VCO is routed through a separate delay.
This achieves much the same goal as polyrhythmic delays, now with harmonies thrown in. With a harmoniser, playing a single note causes a cascade of harmonies to come through with their own rhythmic (or not) timing. Doing this with a modular synth means that you would play a chord and have the notes come through at different times with a rhythmic pattern just as the harmoniser example does.
This example is taken from the same album again, this time from the track Palette 4.
13. Resonator (Karplus-Strong Style)
With very short delay times and very high feedback levels, we start to leave the realm of rhythm and enter that of pitch.
This works well with any style of delay effect, as long as it allows for delay times under 20 milliseconds. Turn the feedback up to 100%, and you will start to hear the delayed sound as an audible pitch. Tapping on the instrument’s pickup (or sending a short pulse another way) will make the delay resonate at its selected frequency.
This is the same way that some synthesizers simulate plucked strings, using Karplus-Strong string modelling – a filtered delay driven with a quick pulse of noise to start the resonance.
When playing a note with the same frequency as the delay, it starts to resonate considerably louder than the input signal, so it’s best to have a limiter set up to avoid clipping or deafening.
I have written a guide at the end of the article with the relationship between delay time and pitch, using the formula t=1/f.
This is a short sample from Palette 1, where the delay time is constantly repeating a sequence of 3ms, 4ms, 5ms and 6ms delay times to give the illusion of an arpeggio.
14. Oscillator (Positive Feedback with a limiter)
Most delays allow for signal to be fed back into the input up to a certain extent to avoid a positive feedback loop (much like the feedback you’d get from a microphone pointed at a speaker).
So what happens if we go overboard and start a positive feedback loop? Usually it becomes hard to control the volume as it gets louder and louder, so once again a limiter of some sort is essential.
The neat thing about this is that we have essentially made an oscillator, which we can control using the delay time in the same way as the resonator example above. Instead of being a simple sawtooth or square waveform, this technique allows us to make a waveform unique to the sound that sets off the resonance cascade (one for the Half Life fans).
While I wouldn’t necessarily use this like a regular instrument, it can make some really out-there sounds that you wouldn’t usually find yourself creating.
Check your headphone levels for this one
15. Binaural Shenanigans (Mid-Side Process via Dan Worrall)
I hinted about mono compatibility in #2 (stereo doubling), as it is always something to consider when you’re not in full control of the listening environment (which is most of the time). Having two identical signals slightly out of time is great in stereo, but in mono the two channels are summed together and cause odd artefacts unless prepared properly. This is where mid-side processing comes in.
If mid-side processing is new to you, the most important part to know is that you can have things sound one way in mono, and another way in stereo, with the same mix.
In Dan Worrall’s excellent video about mixing on headphones and binaural sound, he showed that delaying the mid or side channel can introduce the sensation of 3D sound when using headphones.
In this case, delaying the mid channel by under a millisecond introduce timing differences between your ears much like it occurs naturally – since there is a small distance between our ears, it takes sound slightly longer to reach the furthest ear from the sound source.
When listening to the example below on headphones, the stereo version can sound like it is coming from behind or in front of you, rather than being louder in one ear like we are accustomed to with stereo mixes.
When the mix is played back in mono, the side channels are cancelled out, leaving only the mid channel with an imperceptible delay. No comb filtering, just the original audio.
In case your delay effect doesn’t provide the subdivision options you need (or any at all), here’s how to use maths to find out how long to set the delay time in milliseconds.
Step 1. Use This Handy Website I Found
Step 2. Profit.
Alternatively, you can use a calculator (or your brain if you’re good at mental arithmetic)
Since we commonly use Beats Per Minute to define time, we’re going to start by converting BPM to Hertz (beats per second). That way, everything from here on is based on decimal values.
120BPM divided by 60 seconds gives us two beats per second, or 2Hz.
Time is 1/Hz, or 1 divided by the Hz value we got just then. In this case we have 1/2 or half a second – 500 milliseconds.
So the delay time for a (crotchet) beat at 120BPM is 500ms.
What about two beats? that would be twice as long, so a minim would be one second.
A dotted 8th note would be an 8th note plus a 16th note. An 8th note is half as long as a crotchet at 250ms, and a 16th is half again at 125ms. The total is 375ms for a dotted quaver.
Here is the algorithm for a beat:
time (in msec) = 60,000 / BPM
And for subdivisions:
time (in msec) = 60,000 / BPM / notes per beat
Some quick values you can use:
8th note quintuplet (5 per crotchet) = 12,000 / BPM
16th note (semiquaver) = 15,000 / BPM
8th note triplet (3 per crotchet) = 20,000 / BPM
8th note (quaver) = 30,000 / BPM
Dotted 8th note (dotted quaver) = 45,000 / BPM
Quarter note (crotchet) = 60,000 / BPM
Dotted Quarter Note (dotted crotchet) = 90,000 / BPM
To calculate frequencies:
For resonator and oscillator patches, it’s handy to know how to set the delay time for a particular note (which will be expressed as a frequency).
Since frequency is 1/time (Hz = 1/seconds), then we can quickly get to the formula needed for the delay time:
time (in msec) = 1000/frequency
A4 is 440Hz, so we can use 1000/440 to get 2.27ms for the delay time. Unfortunately, most digital delays only allow for integer values in milliseconds, so we can’t set decimal values.
Here are the first few values using integers with the closest note and tuning:
1ms is B5 + 21 cents (1kHz)
2ms is B4 + 21 cents (500Hz)
3ms is E4 + 19 cents (333Hz)
4ms is B3 + 21 cents (250Hz)
5ms is G3 + 35 cents (200hz)
6ms is E3 + 19 cents (166.6Hz)
7ms is D3 – 47 cents (142.86Hz)
8ms is B2 + 21 cents (125Hz)
9ms is A2 + 17 cents (111.1Hz)
10ms is G2 + 35 cents (100Hz)
11ms is F#2 – 30 cents (90.9Hz)
12ms is E2 + 19 cents (83.3Hz)
13ms is D#2 – 20 cents (76.9Hz)
Funnily enough, this is actually the subharmonic series starting on B5! Who needs a fancy subharmonicon now huh Moog?
I used this website to get the notated pitch and tuning for the above frequency values.