Collaborations of the Third Kind

Perhaps the greatest advantage – as well as the greatest limitation – of composing and producing music using a Digital Audio Workstation (‘DAW’) is that you can create an entire song from the ground up, by yourself, using just a laptop and software in the privacy of your own abode.  For those of us who slant toward introversion this might also rank as its greatest appeal.

But the limitation became obvious to me one day when my paddling friend and guitarist Bill Kuklinski offered to collaborate on a song for ‘Perspectives.”  Never having played a musical instrument I had no experience playing with others or collaborating musically in any way.  So throwing caution to the wind I offered to send Bill a draft of the track “Elements” for his consideration.  “What can you play over this?” I asked, noting that the arrangement was not yet finalized. Could something greater come out of collaboration, especially since we have very different musical backgrounds and overlapping yet distinct tastes in music?

Rather than delve into the particulars of that first collaboration, or even our second collaboration, in this blog post I’ll explore our third collaboration.  Each took a different form: for the first, Bill playing over tracks I had mostly completed; for the second, him sending me three parts of an improvised jam for which I provided the “band” and arrangement; and our latest and perhaps most interesting experiment – sampling.

—

A few weeks ago my e-mail popped up a message from Bill with the intriguing title, “Ambient this!”  He had recorded himself playing the Beatles’ song “She Said She Said,” and wondered if I had any thoughts on where to go with it.  If you’re not familiar with the original song, you can listen to it here.  I highly recommend that you do before we go any further. It will take less than 3 minutes.  

Next, here is Bill’s performance, played on a Gibson semi-hollow bodied jazz guitar.  It is also worth listening to in its entirety because (1) it’s lovely; and (2) you’ll better appreciate what I did with it later.

Obviously, I could once again be the “band” beneath this performance.  Having already done that, and since I saw his “Ambient this!” as a challenge, I thought it would be interesting instead to sample Bill’s performance and use those samples to build a track in an entirely new genre: dub techno.  This turned out to be great fun, and in the process I believe I’ve done a fair job of deconstructing the essential melodic bits of the original song.

There are many ways to sample music.  Just listen to the Beastie Boys’ ‘Paul’s Boutique’ for a masterclass.  You can extract short audio samples from an original performance and drop them into your song as points of emphasis, themes or melodies, or to establish a referential link from your work to the original.  Since I wanted to use the melody from Bill’s performance, and change the tempo from his original 100 beats per minute (‘bpm’) to 110-115 bpm to be stylistically consistent with the dub techno genre, I would have to slice up his performance rather than just sample it.

By “slice,” I mean using a piece of software to find and mark the locations of notes in the original recording. Identifying the locations of transients – what are called slice points - marks the start and end points of notes or percussive elements in an original recording.  You then export a file in .rx2 format (also known as a ‘REX’ file) that comprises the original audio along with its associated slice points and certain metadata.  Software within the DAW can then run sliced notes through algorithms to shorten or lengthen them in order to “fit” the audio into a new tempo, all without changing their tones.  An F# remains an F# even when mapped from 100bpm to 115bpm, and so on.

I used a program called Recycle to slice the main melody from Bill’s guitar recording.  A screen grab appears below.  Notice how notes have a rapid rise time (also called its attack), then a gradual decay which tends to follow an exponential envelope until the next note is struck or the note is muted.  Slice points are the vertical lines marking these attack transients.  The software will make some initial guesses as to where the transients occur.  After that you have to add, delete, or tweak slice point locations.  It’s slow, exacting work.  You also have to tell the software where (for example) the melody starts and stops, the time signature (here, 4/4), and how many measures are in the sample plus any extra beats.  This helps the software compute the original tempo, and thus the time references for slice points.

Bill’s guitar melody, sliced.

You don’t generally import an entire song as a REX file.  Creating that big an .rx2 file would require a lot of tedious work; you most often only want to use a subset anyway.  You might extract the melody and the bridge, and perhaps one or two more parts depending on whether you have a use for them in your own composition.  Besides, if you’re importing the entire song as a REX file you’re just going to be the backing band to the performance; that was not the goal here.

 Here’s what the extracted melody sounds like, now mapped to 110bpm, and passed through a filter that removes much of the bass line Bill was playing. You’ll notice a bit of stuttering in the guitar playback because I hadn’t time-aligned the note slices (or tossed out a few poorly-placed slices), a process that needs to be done by hand

The REX file looks as follows in Reason’s .rx2 file player:

Melody slices vs. time in Reason’s .rx2 player.

The rectangles correspond to the slices of audio, where the left-to-right extent of each is the note (or two-note hammer, or pitch bend) as sounded between the left edge (its start) and the right edge (its finish before the next note is played, not necessarily the ending of the note since it might not be muted).  The vertical dimension orders the slices from first to last as read from the .rx2 file.  This suggests you should be able to move notes around… and you can!  That is one of the more exciting features of sliced samples: the ability to time align notes in case they were played behind or ahead of the beat, and to move them around like pieces on a chess board to create new melodies “played” by the original performer, or to organize a handful of notes into snippets that can be looped.

—

My goal was to integrate Bill’s playing into a dub techno track.  As a genre dub techno tends to be gauzy and bass-heavy, with 100-120bpm tempos and 4/4 time signature along with stripped down arrangements and percussion, plus lots of reverb and delays.  The atmosphere of a track is as important as any other element, if not more so. 

The ‘dub’ part of dub techno is in tribute to Jamaican dub.  Reggae producers like King Tubby would release a 45rpm record with a radio-friendly single on the ‘A’ side, and a dance club friendly ‘dub’ version on the ‘B’ side.  The dub version might use a subset of the lyrics, loop the melody and bass line, and run it all through a lot of reverb and delays.  Jamaican dub emphasized bass and treble, which paired well with the speaker systems found in many Kingston dance halls in the ‘70s and ‘80s. 

 Moritz von Oswald and Mark Ernestus, producing music in Berlin in the ‘90s under the ‘Basic Channel’ moniker, embraced the dub production ethos to create dub techno.  The track that started it all, “Quadrant Dub,” can be found here – I recommend listening to it as calibration piece for the genre.  Note that there’s not a lot of parts to this track, but a fair bit of production varying filters, reverb, and delays on a limited number of samples and notes/chords.  And more than a little audio ‘mud’ for lack of a better term, reflecting the hazy, gauzy, overdriven feel of this production style.

—

My next step was to see if I could organize the melody and other samples into 8- or 16-bar loops of music that somehow fit together, much like I did in composing “One Thought.”  The point in this part of the process is to experiment and create a number of iterations.  Some of these may prove useful; many won’t.  Here are some successive experiments. In the first I merely imported the melody into the DAW (I use Reason), added some delays and reverb, and a placeholder percussion track. You’ll again notice a bit of stuttering in the guitar playback because I still hadn’t time-aligned the note slices.

In the next experiment I extracted a few notes from the melody, rearranging and using them as the modulation input of a vocoder in order to have the guitar notes modulate a chord from a long legato synth pad (you may have heard vocoders used by artists like Pink Floyd; the album ‘Animals’ uses a spoken prayer as the modulation source on the song ‘Sheep’ to create a robotic voice effect). I then ran the vocoder output through a 3-band effects device to add additional ornamentation. Very atmospheric. There’s a placeholder 2-note synth bass line plus a bit of percussion. Yes; Bill’s guitar is in there.

The final snippet pulls the subset of guitar notes out of the vocoder mix a bit so that you can hear the three notes that served as modulation source; they sound like six notes because a delay doubles them. This all sits under the melody, which is buried in the mix. No worries; it’s a draft. I discarded the earlier percussion in favor of a shaker loop with halved tempo. Then about halfway through you’ll note a punchy and distorted synth rise and fade in the mix. You’ll hear a subset of those chords in the second bridge of final mix, but played through a different synth, and used only four times.

As I created each snippet I uploaded it to a Dropbox folder Bill and I share. I reassured him that these were not final arrangements but merely ideas to be considered, dismantled for useful parts, or discarded.

Since dub techno tends not to feature melodies at all, I listened to the (very) few producers who had incorporated vocals into their productions for inspiration.  The biggest takeaway was to leave space – lots of space – in the mix, rather than fill up all the available time with melody.  Melody could be highlighted over certain parts of the song, but it should play nicely with the other elements and not draw too much attention to itself.  This meant I would further strip down the sampled melody to only certain notes, and run it through a couple of filters and delays consistent with the dub techno style.  

The first filter, an ADSR (Attack, Decay, Sustain, Release) would limit the attack of each note, as if the strings were being plucked gently with a soft pick.  A fast attack, like the strike of a snare drum, immediately draws your attention.  With a more gradual build for each note the slower attack both slightly muffles the start of the note, and makes it feel like the playing is slightly behind the beat. This creates a more laid-back feel that can nestle into a mix.  And just for fun I ran the output of the ADSR into a second filter with its center frequency (defining its tonal passband) controlled by a slow oscillator signal running out of sync with the song’s master time reference.  This resulted in a wah-wah effect that I found appealing.  I added a 2-3 ping-pong delay (comprising one 2/8ths of a measure delay panned hard left, and a 3/8ths of a measure delay panned hard right; the two play off each other across the stereo field) plus reverb to provide depth and further interest.  You can listen to the combination of melodic note depopulation and the entire effects chain below:

Most dub techno is played in minor keys; She Said She Said is in the major key of ‘A’, but there you are. Another characteristic of dub techno is the so-called dub stab.  A note or chord is played on a synthesizer – most often a subtractive synth using a saw waveform – having an ADSR filter set to fast attack, decay, and release (e.g., a quick “stab”), and typically an inline chorus effect and more than a fair bit of compression.  Short notes or chords are played in quick succession, leaving mostly space over a measure.  After much experimentation I settled on this. Notice how overdriven it sounds, consistent with the dub techno style:

In order to add movement and space I incorporated a delay with a 5/16 tempo, a huge reverb as a send effect (meaning that in the final mix you hear the original synth combined with the synth’s output ‘sent’ through the reverb), and manually adjusted the synth’s attack and the velocity of the stabs.  Here are the same four measures with the effects chain in place:

Next I extracted a couple of notes from the melody’s REX file and used them as “guitar stab”:

Adding a shorter 1/16 delay than the dub chord stab served to further contrast the guitar from the chords.  Together they do a kind of call-and-response over 8 measures:

I then settled on the general outline of an arrangement – or at least the intro and main theme– which looked like this:

Quick arrangement sketch. But what about the bridge(s)? The outro?

I added a pad with two chords and a slow attack (most notable at the beginning, but present through most of the song in the background), a second pad that created a variable wind-like atmosphere based on its modulation wheel setting, synthesized rain to serve as a bridge into both occurrences of the melody, and finalized the bass line. By this point I had most all the elements.  I just had to pull apart and recombine the pieces, add what was missing (not much), and remove what was extra (quite a bit).  The first partial arrangement showed promise.  I passed it along to Bill; he was encouraging, and asked if I might include the original song’s bridge.  

So back to Recycle I went to sample the bridge.  I only used half of it.  And rather than place it in the song between the two appearances of the main melody I put it in the song’s outro.  I pulled out a few more guitar note subsets, looped them, and ran them though other effects chains – the one in the second bridge is especially notable since it briefly changes the mood of the piece. Add a little percussion, and some more ornamentation, and the track was done.  You can listen to it here:

So what did I get from this exercise?  First, collaborating with someone who is open and receptive to new song forms is, well, fun! And who doesn’t like fun?  I found it possible to compose in a genre of music that I love, and contribute to the corpus by adding melody.  And it was rewarding to see how far I could deconstruct the original melody and bridge, yet still honor the original Beatles composition and Bill’s performance.  I anxiously await his jazz waltz version.

(c) 2020, Shawn Burke, all rights reserved.