Here’s my 23rd free track I wrote within a week.
Aims for this week
What I tried to do differently from what I usually do was:
- Go at a faster tempo, but don’t go genre-y on it – try to be “me”
- C major!
- Use sung vocals (a late addition)
Reckons
One session, maybe 3 hours or so. I’m quietly impressed with myself, but it has been, well just over 33 1/3 years that I’ve been making electronic music.
This was a conscious effort to not do a genre exercise, but instead to try to extend a more Jet Jaguar-y track in a different way. This, even though I agreed with myself (heh) last week that genre exercises are fine. And then I accidentally went kind of shoegaze, kind of post-rock a la Seefeel! That was not the plan, like only after I added the vocal parts and chucked a bunch of delay on them did the other sounds suddenly seem kind of fitting in that way. The pad sound suddenly reminded me of a washed out wall of guitar and the synth bass of a sorta dubby bass guitar with the tops rolled off…
Process
I set my DAW, Bitwig, to 5/4 time, because that’s pretty “me”. I set the bpm to 118, when I usually do stuff between about 80 and 100 bpm.
The first thing I did was create the flutey melody using two modulated arpeggiators, which you hear right at the start of the track. I’ve done this kind of thing a couple of times recently, so thought this was also a good way to not get too genre-y. The guts is to have copies of the same sound panned left and right, and they’re playing a sequence of notes in different timings so the effect sounds like just one line bouncing back and forth in stereo. To start this I just sequenced a big old chord to play constantly for 4 bars (Cmaj9, I guess) and had a built-in synth playing this, just a plain sine tone, no filters or anything. I then activated an arpeggiator and configured this to repeat every 11 steps. I added a little step sequence pattern that modulates the base rate of the arpeggiator so it plays a new note anywhere from once a bar through to once every 16th beat. I made the step sequence some irregular length again, I think it’s 31 steps. Then I added a second sequencer that was a different length again and which only switches between straight and dotted notes.
Then I duplicated the above setup, and hard-panned the two synth lines. I changed the modulation timings and numbers of steps on only one copy, so the arpeggios on each channel move in different rhythms. This gives the result I described above.
I sent both flutey synths into this weird pitch shifter Fermenter I’ve been putting on everything recently, which adds the sludgey, not-just-a-sine-anymore character to the sound, and then into a nice big delay.
I added the most basic one-note synth bass. The same out of the box Bitwig synth, Polymer. It’s literally just knocking out the root note.
I generated some bippy little FM percussion sounds, the kind of conga- and snare-like sounds. They’re a preset that Bitwig-enthusiast Polarity gave away. I recorded some I liked and made a 2-bar pattern with them. I made a variation on the pattern that has an extra hit in each bar.
I found a small-ish kick on my computer and sequenced a 4-bar pattern with some quieter notes set to about 30% chance of actually playing between the main kicks. I bounced out 96 bars of that pattern.
Next up, I copied that one chord that generates all the flutey notes and played it straight, sustained for 4 bars at a time, as a grungey pad sound on an FM synth. I sent this into a big reverb (Valhalla VintageVerb) and put it relatively low in the mix.
The next obvious “me” thing to do was to drag in a semi-random field recording from my folder full of them, this one a recording in Seville cathedral I made in 2014. I created volume automation that ramps up from silence over 12 bars, then cuts dead and sits silent for another 4. Something about the (literal) dynamics of having a mix of slow fades vs. cold stops or starts often makes me happy. The recording is longer than the track I ended up with, so different parts of it fade in every 16 bars. I did absolutely nothing else to it, FX-wise, not even EQ. It’s quite noisy in the mix at times and I liked that.
At this point I arranged the loops I had been working with into a 96-bar long structure. I used a classic low-pass filter to bring in the drums at the start, otherwise it was mostly fading things in and out. I decided having literally only one note on the bass was a bit boring so went absolutely wild and added an occasional D at the end of every 8 bars.
I decided to add a bridge in the relative minor key halfway through the track, made by just moving all the tuned stuff from all sitting around Cmaj9 to Am9. Making it a bridge rather than a coda, I made a return to the original Cmaj9 after 32 bars. Even thought I was happy sitting on the first chord for 48 bars I found the bridge boring so hacked it back to 16 bars, shortening the overall track length. I think it’s an improvement for sure.
In the bridge, I made the flutey sounds switch from sine to triangle waves, which are much brighter and more obviously synth-sounding. I decided they wanted more delay, so turned up the feedback on the delay during this section and also in the final 16 bars after the bridge. (In Bitwig, I achieved this by turning the Polymer instances into Poly Grid – I love this flexibility! – added Triangle and Merge modules and then automated the cross-fade knob on the Merge.)
At this point I had the brainwave to add a woman singing, remembering I had downloaded this free pack of improvised recordings from Freesound. I edited three different files into different variations, playing with the pitch as I went to get them in key, but otherwise letting them play out pretty much as I found them. That said, after listening to them in context I ended up pitching up most of them by a full octave on top of the other tweaking I had done. It’s only the last couple of notes as we come out of the bridge that are close to their original pitch.
I sent the voice into 4 delays of different length that are feeding back into each other, for a big ol’ dubby wash-out. I did some volume automation of the delay tails to extend their life – as the feedback is naturally fading away I crank up the volume of the track.
I wanted to introduce more percussion, so brought in a simple shaker sound and wrote a one-bar pattern. I decided to only include this in the bridge. I did some edits so it doesn’t just repeat over and over, there are more gaps near the end of the bridge.
The last thing I added for some further variation is just a single piano note evenly tapping out 5 beats at the end of every 4 bars. This plays when the main section is kind of getting going, fading up from quite quiet to, IMO, quite obvious leading into the bridge.
Then, for some final variation, I added a selection of delay fx to the FM percussion, and made another step pattern of occasionally switching from dry percussion to each of those. This kind of step sequence that sometimes triggers an event is a very me thing to do, it seems. The pic below is telling you the pattern runs for 43 beats, then reverses direction, and keeps ping-ponging like that. The 43 steps are shown in the box with the blue vertical lines. When it’s just black the signal is dry, when it is blue the percussion is being sent into one of the fx on the right. Because the pattern length doesn’t fit the track’s time signature, it’s unpredictable from listening whether one of the fx will be chosen when a percussion sound is actually playing or whether it’ll just come and go between the notes.
