In this post, I dig into a profound and under-appreciated expressive feature of Ableton Live: warp markers, the “handles” that enable you to grab hold of audio and stretch it precisely. Warp markers have practical applications for getting your grooves sounding the way you want, but they also open up unexpected windows into the nature of musical time itself.
The image below shows an audio clip containing the Amen break. The warp markers are the little yellow arrows across the top. Each one corresponds to the onset of a drum hit.
Think of an audio clip as a rubber band. At its original tempo, the rubber band is not under tension. If you slow the clip down, that stretches the ends of the rubber band apart. If you speed the clip up, that squeezes/scrunches the ends of the rubber band together. Now imagine that the warp markers are thumbtacks sticking through the rubber band into a corkboard. By moving the thumbtacks around, you can speed up and slow down (stretch or squeeze) different parts of the rubber band independently of each other. One drum hit can be early, the next one can be late, and the one after that can be in its original position. Here’s a tutorial showing how it works.
The image below is zoomed in on the first bar of the Amen break. Notice that none of the drum hits (except the first one) fall exactly on the metronomic sixteenth-note grid. They are all a little bit late. This is musically important!
One typical use case for warp markers is to quantize the timing of musical events, that is, to line them up exactly with the grid lines. Here you can see the Amen with each drum hit quantized to the closest sixteenth note.
Quantization can be a big help if you want to tighten up a rhythmically sloppy performance. However, you don’t actually want to do this to the Amen break. The deviations from the grid aren’t sloppiness, they are participatory discrepancies, the expressive microtiming shifts that give the break its distinctive feel. Quantizing the break sucks the groove out of it.
Anyway, the real fun of warp markers is that you don’t have to use them to place events in their “correct” locations; you can also use them to re-time your audio in radical ways. For example, you could quantize the Amen break to eighth note triplets to give it a jaunty shuffle:
Or you could just drag the warp markers around to random places and see what happens:
In the track below, I used warp markers to alter the timing of the Amen break in various ways, some subtle, some not.
Here’s a detailed guide to what you’re hearing. Each loop plays twice:
- 0:00 – The original break.
- 0:15 – The break quantized to the closest sixteenth note.
- 0:30 – The break quantized to the closest eighth note.
- 0:46 – Quantized to the closest quarter note.
- 1:01 – Quantized to the closest eighth note triplet – this one plays four times because it sounds so cool.
- 1:32 – The first bar of the break stretched (slowed down) a little, the second bar squeezed (sped up) a little, the third bar stretched a little, the fourth bar squeezed a little.
- 1:47 – The first bar squeezed a little, the second bar stretched a little, the third bar squeezed a little, the fourth bar stretched a little.
- 2:02 – Alternately stretching and squeezing every half note.
- 2:18 – Alternately stretching and squeezing every quarter note.
- 2:33 – Alternately stretching and squeezing every eighth note, which creates a cool lopsided shuffle feel.
- 2:48 – Alternately squeezing and stretching every eighth note, which creates a kind of inverse lopsided shuffle.
- 3:04 – The break gradually slowing down to a halt – I tried to do this with warp markers but ultimately cheated and automated the session tempo instead.
I also included some Star Trek samples, because why not:
As you listen to my track, you will notice that the sped-up segments of the break are pitched higher, and the slowed-down segments are pitched lower. This is because the track uses Live’s re-pitch mode for warping the clips. In the analog era (and in the early digital era), if you wanted to change the speed of a recording, you also had to change its pitch, and vice versa. If you slow a recording down, that stretches out the waves longer, which is the same as decreasing their frequency and thus lowering their pitch. If you speed a recording up, that makes the waves shorter, which is the same as increasing their frequency and thus raising their pitch. More recently, digital audio engineers have developed methods for changing the pitch and speed of digital audio independently of each other through the magic of phase vocoding.
All of Live’s warp modes other than re-pitch use various flavors of phase vocoding. I chose to use re-pitch in my track because the pitch changes make the time changes more obvious. Beyond the pedagogical value, it also just sounds good.
Here is another warp marker experiment using Duke Ellington’s “Half The Fun (Lately)“, which has one of my favorite drum patterns of all time.
In this track, I used Live’s Beats mode, which keeps transients intact and then fills the space until the next transient with copies of short segments of audio. This creates a cool stuttering effect when you stretch things too far.
It has been possible to correct or alter the timing of a recording since the advent of digital audio editing. You just slice up the audio file into individual drum hits or notes and move them around as you see fit. However, in a program like Pro Tools, this process is tedious and labor-intensive. You would do it to correct timing mistakes in a professional recording, but you probably wouldn’t want to do it for fun or as an experiment. Live’s warping functionality is a big deal because it makes re-timing audio so easy, and because you can hear the results of your experiments in real time rather than after a lag. Editing audio in Pro Tools feels like editing a Word document, but moving warp markers around in Live feels more like improvising music. The possibility of playful goofing around creates space for discovery. Here is another warp marker experiment, using samples of various Led Zeppelin songs. This took me less than an hour.
Creating and playing back loops is easy, but creating loops that don’t immediately get boring or annoying is extraordinarily challenging. Microrhythmic nuance matters a lot in electronic music, because if you repeat a loop many times, listeners will pick up on details that they don’t notice on a single pass. Human drummers always have slight fluctuations in their tempo, but you typically aren’t conscious of them. However, if you take a sample of a live drum performance and loop it, then the participatory discrepencies take on structural significance. Good hip-hop producers know that you don’t necessarily want your drum hits exactly on the grid; moving the snares a little early or the kicks a little late might give you that special bounce, that hump.
J Dilla was especially good at turning slightly “wrong” timing into extremely compelling loops. Dan Charnas explains how Dilla used the affordances of the Akai MPC’s swing functionality along with samples of imperfect live performances to create his distinctive time feel.
Dilla liked to layer different seemingly incompatible grooves together: kick drums programmed with heavy swing, hi-hats programmed with lighter swing, snare drums in straight rhythm but a little early, tambourines in straight rhythm but a little late, and so on. Live’s warp markers make it possible to apply this kind of fine-grained expressive timing to audio recordings. I don’t have enough control over my body to be able to play the grooves that I want on an instrument in real time, but if I can record a general approximation, then I can use warp markers to dial in the groove. I can also experiment with them to discover rhythms that I wouldn’t have been able to find any other way.
What does it feel like to listen to rhythms with this kind of fine digital manipulation? In “Pop as Process: The Digitalization of Groove, Form and Time“, Anders Reuter says:
Digital processes flatten the hierarchy and level the ground between the human and its coexistence with the nonhuman… New levels and kinds of complexity are created in the infinitesimal details that are subject to discrete variations between loops and interrelational functions across sound sources. This demands new kinds of attention, skill and talent from the producer. The craft or creativity consists of setting, editing and adjusting interrelational processes through combinations of mainly metadata.
Digital pop music offers very intricate and particular technology-signifying heterogenous temporalities that are based on a negotiation of the relationship between music and time in ways that are innately digital… it is exactly here—in the ambiguity and complexity of the temporal—that the listening pleasure of digitally produced pop music resides.
Beyond re-timing drum breaks to produce electronic music, Live’s warping functionality has also been an unexpectedly valuable tool for my understanding of Western European classical music. Live’s creators intended it to be used for music in (or close to) metronomic time. However, you can warp any performance to the grid, even if it uses extreme rubato; you just have to know where the notes are “supposed” to go.
You can determine how the audio is supposed to map to the grid by reading a score, but for me it’s easier to do it by ear. Most of the classical canon is available on the web in MIDI format. I just download a MIDI file of the piece I’m working on, import it into Live, and then align an audio recording of a performance to it. This can be very labor-intensive, but once it’s done, then making visualizations like the one in the video above is effortless.
I can also change a classical performance’s rhythmic feel and hear how it sounds. For example, I was curious to hear how the Bach Chaconne would sound over dance beats. The chaconne is supposed to be a dance form, but no one ever performs Bach’s Chaconne with danceable rhythm. By warping out various performances, I could experience their rhythms the way they are notated on the page, without performers’ interpretive timing choices. I could also have done this by reading the score and imagining how the notes sound, but I don’t read well enough for that to be practical. Besides, once I had the Chaconne warped to the grid, then I could add beats to it, and experience the pleasure of Bach if he made modern-day dance music. Hearing the Chaconne this way was crucial to my eventually learning how to play it (however slowly and badly.)
Warping is not just useful for making beat-driven remixes, though. Once you have the piece warped out, then you can set it to be the tempo leader. Rather than forcing the performance to conform to the Live session’s metronomic tempo, the Live session automatically adjust its tempo to follow the nuances of the performance. This makes it possible to align beats and MIDI synths to a rubato classical performance. It can sound pretty cool!
Now that I can turn the rubato on and off in these pieces, it makes it possible to hear what exactly rubato adds to the music and what it takes away. Again, this is something I could just imagine for myself if I were a more adept music reader, but being able to actually hear it has been pretty revelatory. It has definitely helped me articulate why rubato bothers me so much of the time. For example: in his rhythmic analysis of Chopin’s Étude op. 10, no. 3, John Rink argues that the piece makes sense only if performers play the rhythms as Chopin wrote them. Rink complains that when performers are too free with the tempo, then they put rhythmic emphases in the expected places, rather than in the unexpected ones that Chopin actually wrote. I was able to demonstrate Rink’s analysis to myself by listening to Maurizio Pollini’s 1990 recording of the Étude without Pollini’s interpretive timing.
When you push a timestretching algorithm beyond its intended limits, it produces audible artifacting, in this case a stuttering sound. The further Pollini departs from metronomic time, the more extremely Live has to stretch the audio to keep it aligned to the grid, and the more prominent the stuttering effect becomes. In other words, you can tell how much rubato Pollini is using in the original recording by listening to how stuttery the audio is in the remix. I’m pretty pleased to have accidentally found a way to shift this expressive dimension of the music from the tempo to the timbre.
After hearing how good some of this music sounds without rubato, I started to wonder where all the rubato came from in the first place. I just could not believe that, like, Bach would have wanted people to play his music with so much tempo fluctuation. Then I read Richard Hudson’s book Stolen Time: The History of Tempo Rubato and his Grove Dictionary entry on rubato, where I learned that rubato as we know it now is a fairly recent development, going back only to the mid-late 19th century. Before that, soloists and singers could be somewhat free with tempo, but accompanists generally kept steady time. Hudson quotes an 18th century treatise on Italian arias: “When the bass goes an exactly regular pace, the other part retards or anticipates in a singular manner, for the sake of expression, but after that returns to its exactness, to be guided by the bass.” This sounds more like the timekeeping in an Ella Fitzgerald recording than in a present-day classical performance.
Before the Romantic era, the timekeeping was probably mostly steady in solo keyboard music too. There might have been some time displacement in the right hand, but Hudson quotes Mozart as saying that “the left hand should go on playing in strict time.” Even Chopin probably liked to hold the tempo together pretty tightly in his left hand. Chopin! People didn’t start going really overboard with rubato until Liszt and Wagner, and even when other Romantic-era composers wrote rubato into their scores, they didn’t necessarily use as much of it as you would think. Debussy made a few recordings in his lifetime, and in them he uses way less rubato than present-day interpreters of his music do. I may not be able to perform any of this music according to my own rhythmic sensibilities, but I can easily get other people’s recordings to sound the way I want, and it’s validating to know that I might be getting closer to the composers’ own intentions that way.
Metronomic time is a strange thing. The metronome was only invented in 1814 and didn’t enter widespread use for many years after that. The Romantic push towards wilder and wilder rubato may have been a reaction against the advent of mechanically perfect time. Beethoven was enthusiastic about the metronome, and he was fascinated by musical automata like barrel organs and cuckoo clocks. Wagner was not enthusiastic about them at all. The mass audience sided with Beethoven; we are all saturated in electronically produced music in perfect clock time, and no longer think that there’s anything strange about it.
I know plenty of musicians who think that it’s impossible to be soulful or expressive using drum machines, MIDI-controlled synthesizers and DAW quantization. I do not agree; Prince and J Dilla had no problem making drum machines swing. Computers just change the nature of the challenge. For centuries, it was a struggle to produce steady and accurate rhythm. But the clock made that easier, and the computer makes it effortless. So now the challenge becomes, how do we make time into something more expressive than just mechanical plodding? One solution is to have performers and conductors speed up and slow down whenever they feel like it. I understand that solution, but I don’t love it, because it inhibits my ability to connect with the music. It becomes about the whims of the performer or conductor, not about a shared participatory experience. Social music needs to be steady so that everybody can predict where it’s going (at least some of the time). But just because social music has to be rigidly predictable at the macro level, that doesn’t mean you can’t have lots of surprise and nuance at the micro level.