Delays, Feedback, and Filters: A Trifecta
Interaural time difference, interaural level difference, and head shadow are useful not only for an audio engineer, but are also important for us when considering the effects and uses of delay in electroacoustic musical contexts.
My last post, “Delays as Music,” was about making music using delays as an instrument, specifically in the case of the live sound processor. I discussed bit about how delays work and are constructed technically, how they have been used in the past, a bit about how we perceive sound, and how we perceive different delay times when used with sounds of various lengths. This post is a continuation of that discussion. (So please do read last week’s post first!)
I wrote about our responsiveness to miniscule differences in time, volume, and timbre between the sounds arriving in our ears, which is our skill set as humans for localizing sounds—how we use our ears to navigate our environment. Sound travels at approximately 1,125 feet per second but though all sound waves we hear in a sound are travelling at the same speed, the low frequency waves (which are longer) tend to bend and wrap around objects, while high frequencies are absorbed or bounce off of objects in our environment. We are sensitive to delay times as short as a millisecond or less, as related to the size of our heads and the physical distance between our ears. We are able to detect tiny differences in volume between the ear that is closer to a sound source and the other. We are able to discern small differences in timbre, too, as some high frequency sounds are literally blocked by our heads. (To notice this phenomena in action, cover your left ear with your hand and with your free hand, rustle your fingers first in the uncovered ear and then in the covered one. Notice what is missing.)
These psychoacoustic phenomena (interaural time difference, interaural level difference, and head shadow) are useful not only for an audio engineer, but are also important for us when considering the effects and uses of delay in electroacoustic musical contexts.
My “aesthetics of delay” are similar to what audio engineers use, as rule of thumb, for using delay as an audio effect, or to add spatialization. The difference in my approach is that I want to find a way to recognize and find sounds I can put into a delay, so that I can predict what will happen to them in real time as I am playing with various parameter settings. I use the changes in delay times as a tool to create and control rhythm, texture, and timbral changes. I’ve tried to develop a kind of electronic musicianship, which incorporates acousmatic listening and quick responses, and hope to share some of this.
As I wrote, it’s all about the overlap of sound. If a copy of a sound, delayed by 1-10ms, is played with the original, we simply hear it as a unified sound, changed in timbre. Short delayed sounds nearly always overlap. Longer delays might create rhythms or patterns; medium length delays might create textures or resonance. It depends on the length of the sound going into the delay, and what that length is with respect to the length of the delay.
This post will cover more ground about delays and how they can be used to play dynamic, gestural, improvised electroacoustic music. We also will look at the relationship between delays and filtering, and in the next and last post I’ll go more deeply into filtering as a musical expression and how to listen and be heard in that context.
Mostly, I’ll focus on the case of the live processor who is using someone else’s sound or a sound that cannot be completely foreseen (and not always using acoustic instruments as a source– Joshua Fried does this beautifully with sampling/processing live radio in his Radio Wonderland project). However, despite this focus, I am optimistic that this information will also useful to solo instrumentalists using electronics on their own sound as well as to composers wanting to build improvisational systems into their work.
No real tips and tricks here (well maybe a few), but I do hope to communicate some ideas I have about how to think about effects and live audio manipulation in a way that outlasts current technologies. Though some of the examples below will use the Max programming language, it is because it is my main programming environment, but also well suited to diagram and explain my points.
We want more than one, we want more than one, we want…
As I wrote last week, musicians often want to be able to play more than one delayed sound, or to repeat that delayed sound several times. To do this, we either use more delays, or we use feedback to route a portion of our output back into the input.
When using feedback to create many delays, we route a portion of our output back into the input of the delay. By routing only some of the sound (not 100%), the repeated sound is a little quieter each time and eventually the sound dies out in decaying echoes. If our feedback level is high, the sound may recirculate for a while in an almost endless repeat, and might even overload/clip if we add new sounds (like a too full fountain).
Using multi-tap delays, or a few delays in parallel, we can make many copies of the sound from the same input, and play them simultaneously. We could set up different delay lengths with odd spacings, and if the delays are longer than the sound we put in, we might get some fun rhythmic complexity (and polyrhythmic echoes). With very short delays, we’ll get a filtered sound from the multiple copies being played nearly simultaneously.
Any of these delayed signals (taps) could in turn be sent back into the multi-tap delay’s input in a feedback network. It is possible to put any number and combination of additional delays and filter in the feedback loop as well, and these complex designs are what make the difference between all the flavors of delay types that are commonly used.
It doesn’t matter how we choose to create our multiple delays. If the delays are longer than the sounds going into them, then we don’t get overlap, and we’ll hear a rhythm or pattern. If the delays are medium length (compared to our input sound), we’ll hear some texture or internal rhythms or something undulating. If the delays are very short, we get filtering and resonance.
The overlap is what determines the musical potential for what we will get out of our delay. For live sound processing in improvised music, it is critical to listen analytically (acousmatically) to the live sound source we are processing. Based on what we hear, it is possible to make real-time decisions about what comes next and know exactly what we will get out.
Time varying delay – interpolating delay lines
Most cheaper delay pedals and many plugins make unwanted noise when the delay times are changed while a sound is playing. Usually described as clicks, pops, crackling or “zipper noise”, these sounds occur because the delays are “non-interpolating.” These sounds happen because the changes in the delay times are not smooth, causing the audio to be played back with abrupt changes in volume. If you never change delay times during performance, fixed simple delays and a non-interpolating delay is fine.
Changing delay times is very useful for improvisation and turning delay into an instrument. To avoid the noise and clicks we need to use “interpolating” delays, which might mean a slightly more expensive pedal or plugin or a little more programming. As performers or users of commercial gear we may not be privy to all the different techniques being used in every piece of technology we encounter. (Linear or higher order interpolation, windowing/overlap, and selection of delayed sounds from several parallel delay lines are a few techniques.) For the live sound processor / improviser what matters is: Can I change my delay times live? What artifacts are introduced when I change it? Are they musically useful to me? (Sometimes we like glitches, too.)
Doppler shift! Making delays fun.
An interesting feature/artifact of interpolating delays is the characteristic pitch shift that many of them make. This pitch shift is similar to how the Doppler shift phenomenon works.
A stationary sound source normally sends out sound waves in all directions around itself, at the speed of sound. If that sound source starts to move toward a stationary listener (or if the listener moves toward the sound), the successive wave fronts start getting compressed in time and hit the listener’s ears with greater frequency. Due to the relative motion of the sound source to the listener, the sound’s frequency has in effect been raised. If the sound source instead moves away from the listener, the opposite holds true: the wave fronts are encountered at a slower rate than previously, and the pitch seems to have been lowered. [Moore, 1990]
OK, but in plainer English: When a car drives past you on the street or highway, you hear the sound go up in pitch as it approaches, and as it passes, it goes back down. This is the Doppler Effect. The soundwaves travel at the same speed always, but they are coming from an object that is moving so their frequency goes up and then goes down when it is moving again away from you.
A sound we put into a delay line (software / pedal / tape loop) is like a recording. If you play it back faster, the pitch goes higher as the sound waves hit your ears in faster succession, and if you slow it down, it plays back lower. Just like what happens to the sound of a passing siren from a train or car horn that gets higher as it approaches and passes you: when delayed sounds are varied in time, the same auditory illusion is created. The pitch goes down as delay time is increased up as delay time is decreased, with the same Doppler Effect as the case of the stationary listener and moving sound source.
Using a Doppler Effect makes the delay more of an “instrument” because it’s possible to repeat the sound and also alter it. In my last post I discussed many types of reflections and repetitions in the visual arts, some exact and natural and others more abstract and transformed as reflections. Being able to alter the repetition of a sound in this way is of key importance to me. Adding additional effects in with the delays is important for building a sound that is musically identifiable as separate from that of the musician I use as my source.
Using classic electroacoustic methods for transforming sounds, we can create new structures and gestures out of a live sound source. Methods such as pitch-shifting, speeding sounds up or slowing them down, or a number of filtering techniques, work better if we also use delays and time displacement as a way to distinguish these elements from the source sounds.
Many types of delay and effects plugins and pedals on the market are based on simple combinations of the principal parameters I have been outlining (e.g. how much feedback, how short a delay, how it is routed). For example, Ping Pong Delay delays a signal 50-100ms or more and alternates sending it back and forth between the left and right channels, sometimes with high feedback so it goes on for a while. Flutter Echo is very similar to the Ping Pong Delay, but with shorter delay times to cause more filtering to occur—an acoustic effect that is sometimes found in a very live sounding public spaces. Slapback Echo has a longer delay time (75ms or more) with no feedback.
FREEZE! Infinite Delay and Looping
Some delay devices will let us hold a sample indefinitely in the delay. We can loop a sound and “freeze” it, adding additional sounds sometime later if we choose. The layer cake of loops built up lends itself to an easy kind of improvisation which can be very beautiful.
Looping with infinite delay is used by an entire catalog of genres and musical scenes from noise to folk music to contemporary classical. The past few years especially, it’s been all over YouTube and elsewhere online thanks to apps and applications like Ableton Live and hardware like Line 6, a popular 6-channel looper pedal. Engaging in a form of live-composing/production, musicians generate textures and motifs, constructing them into entire arrangements, often based upon the sound of one instrument, in many tracks, all played live and in the moment. In terms of popular electronic music practice, looping and grid interfaces seem to be the most salient and popularly-used paradigms for performance and interface since the late 2000s.
Looping music is often about building up an entire arrangement, from scratch, and with no sounds heard that are not first played by the instrumentalist, live, before their repetition (the sound of which is possibly slightly different and mediated by being heard over speakers).
With live sound processing, we use loops, too, of course. The moment I start to loop a sound “infinitely,” I am, theoretically, no longer working with live sound processing, but I am processing something that happened in the past—this is sometimes called “live sampling” and we could quibble about the differences. To make dynamic live-looping for improvised music, whether done by sampling/looping other musicians, or by processing one’s own sound, it is essential to be flexible and be able/willing to change the loops in some way, perhaps quickly, and to make alterations to the audio recorded in real-time. These alterations can be a significant part of the expressiveness of the sound.
For me, the most important part of working with long delays (or infinite loops) is that I be able to create and control rhythms with those delays. I need to lock-in (synchronize) my delay times while I play. Usually I do this manually, by listening, and then using a Tap Tempo patch I wrote (which is what I’ll do when I perform this weekend at Spectrum as part of Nick Didkovsky’s Deviant Voices Festival on October 21 at Spectrum and the following day with Ras Moshe as part of the Quarry Improvised Music Series at Triskelion Arts).
Short delays are mostly about resonance. In my next and final post, I will talk more about filters and resonance, why using them together with delay is important, as well as strategies for how to be heard when live processing acoustic sound in an improvisation.
In closing, here is an example from What is it Like to be a Bat? my digital chamber punk trio with Kitty Brazelton (active 1996-2009 and which continues in spirit). In one piece, I turned the feedback up on my delay as high as I could get away with (nearly causing microphones and sound system to feedback too), then yelled “Ha!” into my microphone, and set off sequence of extreme delay changes with an interpolating delay in a timing we liked. Joined by drummer Danny Tunick, who wrote a part to go with it, we’d repeat this sequence four times, each time louder, noisier, different but somehow repeatable at each performance. It became a central theme in that piece, and was recorded as the track “Batch 4” part of our She Said – She Said, “Can You Sing Sermonette With Me?” on the Bat CD for Tzadik label.
Some recommended further reading and listening
Thom Holmes, Electronic and Experimental Music (Routledge, 2016)
Jennie Gottschalk, Experimental Music Since 1970 (Bloomsbury Academic, 2016)
Geoff Smith, “Creating and Using Custom Delay Effects” (for the website Sound on Sound, May 2012) Smith writes: “If I had to pick a single desert island effect, it would be delay. Why? Well, delay isn’t only an effect in itself; it’s also one of the basic building blocks for many other effects, including reverb, chorus and flanging — and that makes it massively versatile.”
He also includes many good recipes and examples of different delay configurations.
Phil Taylor, “History of Delay” (written for the website for Effectrode pedals)
Daniel Steinhardt and Mick Taylor, “Delay Basics: Uses, Misuses & Why Quick Delay Times Are Awesome” (from their YouTube channel, That Pedal Show)