Here’s my 28th free track, started and finished within a week, for this year. Over an hour and a half of music so far, pretty nuts. Outside of WeeklyBeats, I’ve finished 10 other tracks and have another 4 on the go at the moment.
Aims this week
This week the things I tried to do differently to usual were:
- Go nuts with spectral processing
- Process the master bus
Reckons
I was pretty inspired by not just a video that I mention below but also the latest False Aralia release, an album-length release as Topdown Dialectic. The “I.S.” person behind everything on False Aralia consistently creates this fucking great tamped-down, reduced sound. I think of it as definitely having roots in the dub techno of Basic Channel and Chain Reaction artists, but it is really different to where most take that sound. Very cool IMO.
Anyway, reckons about my track? I enjoyed doing this one and am quite happy with the result. I could imagine re-using multiple ideas and figured out a simple way to get the delta signal from any plugin or chain of effects in my DAW, Bitwig.
Process
So, like a bunch of music I make I actually started out from a little formal experiment. In this case I took inspiration from watching the promo video for a plugin called Anina, which shows an idea of how to take a certain kind of plugin and (mis)using it a spectral vocoder. So, I made myself a spectral vocoder using a plugin I already have, Trackspacer. I wanted what I’d call the delta output of the plugin – only the effected stuff – and found it was simple to do this in Bitwig by mixing Trackspacer’s wet output with an inverted copy of the dry input. The result sounds heaps like a vocoder, but behaves differently when you play with the parameters of the plugin. More at that video link if you’re curious.
I sequenced an 8-bar chord loop on a simple synth pad sound. I drew in some per-note pitch bends to move between some of the chords, for fun.
For the vocoder sidechain input / modulation source I grabbed some preset drum machine loop from Bitwig’s built-in content. Then I played with Trackspacer settings a lot to see how they changed the results. I set up different LFOs to modulate the high cut, low cut and attack at different rates. For each one I then added a second LFO to gently modulate the speed of the first. None of the LFOs are synced to bars and beats of the project’s tempo in the first place, but modulating the LFO speed really makes the results unpredictable – but it always sounds the same each time I press play.
I processed the vocoder output further, sending it into a copy of the free Spectral Shift plugin to pitch it down an octave. I used yet another LFO to modulate the formants’ pitches via this plugin, so the sound kind of gets brighter and duller. I then sent this into fairly quiet spring reverb (Twangstrom).
I thought it might also be interesting to change the drum sounds’ pitches and see how that changed the way the vocoder reacted. I bounced the drum loop to 4 bars of audio and loaded that into a sampler. I set the sampler to a spectral stretch mode where I could change the pitch without changing the speed of the loop. I chucked another LFO on the pitch of this sample and gave that a listen.
I had planned to only use the drums to shape the chords via the vocoder, but I liked where things were going and decided to include them in the mix. I made sure the dry drum loop was being sent to the vocoder, then started doing more processing of the drums that you hear in the mix. I added some saturation and a low-pass filter with yet another big LFO sweeping its cutoff, and sent this into another spring reverb (Twangstrom again). I let the cutoff frequency drop low enough that the drums practically vanish from the mix.
I wrote a quick bassline, and used a Bitwig feature where you can have a note not play every time a loop repeats. It’s represented by these little blocks inside each note on the piano roll.
So although this is only a 1-bar sequence on the piano roll, that second C is switching on and off in a 7-bar pattern, D in a 5-bar pattern, and B flat in a 6-bar pattern. This creates heaps of variety.
I added a kick drum sound here, but ended up pitching it up so it’s quite bippy and non-doofy.
I sent all the drum sounds into another copy of Spectral Shift, doing the same formant-shifting malarkey as I applied to the vocoder, but at a different rate. I didn’t pitch the sounds down either, just played with formants.
I put yet another Spectral Shift on the bass and a milder amount of formant-shifting.
For the millionth time I used that free sample pack of a woman singing, loading one phrase into a sampler. I experimented with spectral repitching of the sample’s individual harmonics, locking them into the track’s key so it wasn’t too enharmonic and brutal sounding. I added a high-pass filter and modulated the cutoff frequency so, as with the drums, sometimes the sound basically vanishes from the mix. I sent the result into a short, high-feedback delay. The outcome is pretty damn unrecognisable as a voice, IMO. In the final track you first hear it about 1:35.
Next I made short chord stabs, more like a reggae organ skank, duplicating and editing the chord sequence from the vocoder part. This is just a very basic synth sound. To stay true to the spectral processing aim, I sent this into a Bitwig device called Loud Split, which analyses a sound and splits it into 3 bands based on the loudness of the frequencies at any given point in time. I massively boosted the quiet part, muted the middle part, so you’re left with just the loudest bits and the noisy, quiet stuff. This fires off into a dotted delay.
I put a copy of that Loud Split device on the master channel and played around with filtering out the middle part and leaving the loud and quiet bits. This creates a really scooped-out effect that is pretty fun. I put this on the master back in week 9, Slaw, and enjoyed the results.
Finally, at this point, I moved from just a bunch of loops to structuring out a full track. Because of the massive amounts of modulation and things moving around by design, I did almost no edits or even fades in or out. I did automate the above master loudness splitting, so the track starts with everything running into that splitter and fades to a dry signal by about 1:30. The kick drum fades in at the start. Otherwise, I just plonked blocks of sound on a timeline and gave it a listen.