Dub Track — Roots Version Mixed Live Deconstruction
A 72-bar archetypal 1970s roots dub version at 72 BPM in E minor. A one-drop riddim arrives as an eight-channel “tape”, drum and bass first. Instruments come and go on mutes, single words and horn phrases are thrown into a three-head tape echo and a spring, a stepped high-pass “Big Knob” thins the riddim for the drop, and the whole mix is played on faders and recorded into Arrangement as a performance.
- Difficulty
- Beginner
- Live edition
- Standard and up
- Max for Live
- Not needed
The blueprint is an archetype, not a real record. It is the B-side version of an imagined 1970s roots vocal tune: 72 bars at 72 BPM in E minor, exactly 4 minutes. The engineer did not write it. A producer brought in a finished riddim tape, and the engineer mixed it on the desk, live, in one pass. The Dub reggae genre page teaches how each sound is made, and Dub FX & delays and the Dub delay send rack teach the effects. This page covers what those pages do not: how the parts are laid out as desk channels, when each one is open, what the effects do bar by bar, and how to record the mix as a performance. Three decisions carry it. (1) The riddim is fixed and the mix is the composition. Eight audio channels play the full 72 bars without a single edit, and every change you hear is a fader, a mute, a send or a filter. (2) Drum and bass are the only constants. Everything else is a guest that comes in for a few bars, gets thrown into the echo, and leaves. The two drops (bar 29 and bar 49) take away different things, so the record has two shapes of absence. (3) The high-pass filter is stepped, not swept. The page builds a ten-position switched high-pass from a Chain Selector, so the riddim thins in audible clicks the way a passive console filter does, and the bass slams back on one downbeat.
- Tempo
- 72 BPM, counted in one-drop time (snare and kick on beat 3). 1 bar = 3333 ms; 1 beat = 833 ms; 1/8 = 417 ms; 1/8 triplet = 278 ms; 1/16 = 208 ms; 1/32 = 104 ms; 3/16 (dotted 1/8) = 625 ms; 1/4 dotted = 3/8 = 1250 ms; 4 bars = 13.3 s; 8 bars = 26.7 s
- Key
- E minor, two chords: Em (i) and D (bVII). Bass on E0 41.2 Hz to E1 82.4 Hz (Live names, C3 = middle C), with G0 49.0, A0 55.0, B0 61.7 and D1 73.4 Hz. Skank voiced around B2–G3 (247–392 Hz)
- Length
- 72 bars × 3.333 s = 240 s = 4:00. The first drop starts at 1:33 (bar 29), the high-pass drop at 2:40 (bar 49), the outro at 3:33 (bar 65)
- Swing
- The hats play 1/8 triplets (a shuffle), and everything else plays straight 1/16s that sit behind them. Rim +8 ms late, bass −5 ms early, set with Track Delay
- Loudness
- About −12 LUFS integrated, −1 dBTP. The drops are meant to be quieter, so do not limit them back up
- Mood
- Heavy, patient, spacious. Something leaves every four bars, and something comes back as an echo
Where the archetype comes from
The track is invented, so the arrangement, levels and devices below are the page’s recipe. The working method it copies is documented:
- Documented: dub grew out of the instrumental “version” on the B-side of a Jamaican 45. It works by partly or fully removing the vocal and other instruments to bring out the drum-and-bass riddim, adding echo, reverb and delay, and letting parts drop in and out. Larkin is cited for the view that King Tubby and Lee Perry treated the console as an instrument (Wikipedia, “Dub music”, citing Toop, Larkin and Veal). That is the page’s premise: the riddim is fixed and the mix is the performance.
- Documented: independent producers brought their unvoiced rhythm tracks to Tubby’s studio and had their dub versions mixed there. The studio was never big enough to record a full band. In 1971 Bunny Lee helped Tubby get an obsolete MCI desk from Dynamic Sounds, and David Katz names its high-pass filter as an important part of the sound. On “King Tubby Meets Rockers Uptown”, Katz notes that only fragments of Jacob Miller’s vocal are left over the drums and bass (FACT, David Katz, 2015). Hence a finished riddim “tape”, a mix-only session, and a vocal heard as fragments.
- Documented: by 1973 Tubby was working on a modified four-track board. The high-pass filter, a large knob he called the “Big Knob”, was swept across a signal until it narrowed to a thin squeal (Wikipedia, “King Tubby”). A Sound On Sound news piece on a hardware replica says the filter was an Altec 9069B, a big red knob on the MCI. It adds that the filter’s character came from its discrete steps and the clicks they produced (Sound On Sound, 2022). That is why this page’s filter is stepped. The replica is a commercial product, so treat the step detail as a manufacturer’s claim that SOS reported.
- Documented: Scientist, who started at Tubby’s, says it was a four-track machine there. Asked whether the sound came from phasers, he answered, “No, no phasers. High-pass filter.” He says the work was usually one take, because a session had to be recorded and mixed inside the studio time someone had booked (Tape Op #136, interview with Hopeton “Scientist” Brown). This is where the one-pass recording workflow in recording the mix comes from.
- Documented: Roland’s RE-201 Space Echo (1974) has one record head and three playback heads, chosen with a rotary selector, and a built-in spring reverb. Lee Perry used it heavily in the 1970s (Wikipedia, “Roland RE-201”). The Black Ark had a four-track Soundcraft desk, a TEAC tape machine, an Echoplex and a Space Echo (Wikipedia, “Black Ark Studios”, citing Bradley). The three-head echo below is built on that head-selector idea.
- Documented: in the one drop, kick and snare land together on beat 3 and beat 1 is left empty. Carlton Barrett popularised it, though who invented it is disputed (Wikipedia, “One drop rhythm”).
- Documented: Augustus Pablo is the musician most associated with melodica in dub. “King Tubby Meets Rockers Uptown” (1976) is Tubby’s dub of a Pablo production (Wikipedia, “Augustus Pablo”).
- Documented (Live): Arrangement automation is recorded by turning on Automation Arm and Arrangement Record, then moving the controls. The Track Activator (mute) can be automated, and moving an automated control by hand overrides it until you press Re-Enable Automation (Live 12 manual, Automation and Editing Envelopes).
- By ear: these are conventions heard across 1970s roots versions, not claims about any named record: the drum-and-bass entry, guests lasting about four bars, a drop that leaves only bass and an echo tail, horn and vocal lines heard once dry and then only as echo, and the spring crash on the last snare of a phrase. The 72-bar shape and every level and device setting are the page’s recipe.
Arrangement
Page recipe. Bars here are 3.333 s long, so a four-bar phrase is 13.3 s, about as long as a listener will wait for something to change. The intro (bars 1–4) is the “song” for a moment: a two-bar horn fanfare and a dry piano line, then a tom fill with a spring crash. Drum and bass (5–12) is the riddim stripped to its two constants, with the hats easing in from bar 7. The riddim section (13–20) opens the guitar skank and the organ bubble, and the piano doubles the skank from bar 15. In voice (21–28) one line of the vocal side appears, gets thrown into the echo, and its last word comes back at bar 27 as the hats start dropping out. Drop 1 (29–32) mutes everything except the bass. The echo holds whatever was thrown last and is pushed into runaway, and the bass already thinned through bars 27–28 comes back full on bar 29. In bass and rim (33–40) the drums return on the rim alone, then climb back through the high-pass. Horns and melodica (41–48) brings two new guests, each thrown on its last note. The HP drop (49–52) steps the whole riddim, bass included, up through the Big Knob until only a thin squeal and the echo are left. The return (53–64) slams everything back on bar 53. The outro (65–72) is drum and bass again, with one last thrown skank and a word.
Energy curve
Page recipe. In dub the energy curve runs against the wet curve. Each time channels close, the returns take over, so a drop is quiet but not empty. The return at bar 53 is the peak, because it is the only moment when every channel opens together, after the longest stretch of absence (bars 49–52, with the riddim filtered away). Two drops built differently also stop the version from feeling like one trick played twice. Drop 1 is a mute drop. The HP drop is a filter drop.
Transitions
Three bar-level moves carry the track. Each grid shows the last bar before the change and the first bar after it, as 32 steps of 208 ms.
Frequency layering
Page recipe. The genre page gives the static EQ pockets (Dub reggae, mix targets). What the arrangement adds is time-sharing. Skank and organ share 300 Hz–2.5 kHz but leave together at bar 27, and the vocal fragment then has that band alone. Horns (bars 41–44) and melodica (bars 45–48) take turns. The two returns are band-limited, at 280 Hz–3 kHz for the echo and 300 Hz–5 kHz for the spring, so even at 90% wet share in a drop the bass stays clear and the hats keep their air.
Element by element
The riddim “tape”: eight desk channels, bounced once, never edited
Quick start
- Load
- Saturator, Channel EQ, Auto Filter
- Set
- Riddim all parts programmed for all 72 bars, no arrangement in the MIDI
- Bounce Track in Place every part to audio; drums on one stem, hats on another
- Saturator Analog Clip at 1–2 dB per stem, then nothing is re-recorded
- Desk channels eight, each with Channel EQ and an Auto Filter HP, grouped as “Riddim”
- Returns A Echo and B Spring, outside the group
- You should hear
- A finished riddim on eight channels where from now on only the mixer and the returns make the version.
- Live edition
- Intro and up
Write and record the riddim first, all 72 bars, as if a band had played it. Then bounce each part to audio and stop composing. From that point the only tools allowed are the mixer and the returns, which is the discipline that makes it a version and not a remix. Eight channels is the page’s choice, a middle ground between Tubby’s four-track and a later 16-track room.
| Desk ch. | Contents | Sends | Fader at rest |
|---|---|---|---|
| 1 Drums | Kick, rim, snare, toms (one stereo stem) | Spring only (rim, snare) | −13 dBFS peak |
| 2 Hats | Closed and open hat, shaker | None | −22 dBFS peak |
| 3 Bass | Bass (mono) | Never | −10 dBFS peak |
| 4 Guitar | Skank chop | Echo (throws), spring | −20 dBFS peak |
| 5 Organ | Bubble | Echo (rare) | −24 dBFS peak |
| 6 Piano | Offbeat double + intro line | Echo, spring | −22 dBFS peak |
| 7 Horns / melodica | Fanfare, horn line, melodica (shared channel, never together) | Echo, spring | −18 dBFS peak |
| 8 Voice | The vocal side’s lead, uncut | Echo (pre-fader), spring | −inf (opened for fragments) |
- Program the riddim in MIDI for all 72 bars, with no arrangement in it: every part plays the whole length except the intro horn fanfare and piano line, the horn line in bars 41–44, the melodica in bars 45–48 and the vocal. In a real session those would be on the tape all the way through, so record them in full and let the mutes choose.
- Bounce every part to audio (Live 12.2 Bounce Track in Place). Put the drums on one stem and the hats on another, so a hat drop-out is a mute and not an edit.
- Colour each stem once, the way the tape would have been: a Saturator Analog Clip at 1–2 dB per stem, or the print-to-tape chain. From here on nothing is re-recorded.
- Group the eight channels as “Riddim”. Returns A and B and the master bus sit outside the group.
One-drop drums on two channels: the kit, the shuffle hats and the fill into each phrase
Quick start
- Load
- Drum Rack, Drum Sampler, Channel EQ, EQ Eight
- Set
- Kick Drum Sampler soft kick, Decay 350 ms, LP 4 kHz; Channel EQ Low +2 dB, High −3 dB
- Rim cross-stick, Decay 110 ms; EQ Eight HP 180 Hz, bell +2 dB 2.5 kHz
- Layout kick and rim together on beat 3
- Hats grid 1/8T, accents on beats 1 and 3, second triplet of each beat empty
- Track Delay rim +8 ms, kick 0; tom fill in the last half-bar of every four
- You should hear
- A one-drop with the shuffle coming from the hats, a slightly dragging rim and a fill marking every four-bar phrase.
- Live edition
- Standard and up
The sounds come from one-drop and steppers drums. What this track adds is the layout: kick and rim together on beat 3, hats in 1/8 triplets so the shuffle comes from the hats and not from a groove setting, and a tom fill in the last half-bar of every four-bar phrase, which is where the spring crash lands.
- Clip grid at 1/8T for the hats. Accent beats 1 and 3, and keep the second triplet of each beat empty. That gap is the shuffle.
- Rim Track Delay +8 ms, kick 0. The late rim against a quantised bass is the drag the genre page describes, set once here.
- Program the fill bar as above every four bars and vary only its last two hits. It is the phrase marker for the mixing engineer as much as for the listener.
The bass: the one channel that never leaves, and how the desk treats it in each section
Quick start
- Load
- Analog, Channel EQ, Compressor, Utility
- Set
- Bass stem Analog or DI, bounced
- Channel EQ 80 Hz HP off, Low +2 dB in drops, Mid −2 dB 400 Hz
- Compressor 3:1, attack 20 ms, 2–3 dB
- Utility Mono
- Sends never sent to a return; only muted in the HP drop
- You should hear
- The floor of the version, never leaving, thinner under the guests and full and pushed in the drop.
- Live edition
- Standard and up
Build the tone as in Dub bass. In the version its job is to be the floor. It is never sent to a return and it is only muted in the HP drop. What changes from section to section is the desk channel EQ: a little thinner while the guests are in, full and pushed in the drop, and filtered away by the riddim HP at bars 49–52 so its return on bar 53 lands hard.
Guitar skank: in for 14 bars, then only ever heard as a throw
Quick start
- Load
- Amp, Cabinet, Channel EQ, Compressor
- Set
- Guitar DI muted Em / D chops, bounced
- Amp Clean, Gain 2, Bass 3, Middle 6, Treble 6; Cabinet 1x12, Dynamic, Near On-Axis
- Channel EQ 80 Hz HP on, Mid −2 dB 900 Hz
- Compressor 4:1, attack 1 ms, 4 dB
- Send A Echo 0 at rest, 0 dB for one chop
- You should hear
- A clean skank that plays for 14 bars, then returns only as single chops thrown into the echo.
- Live edition
- Suite
The skank recipe is on the genre page (Skank). In this version it is a full member for bars 13–26 and a ghost after that. In bars 41–44 and the outro the channel stays muted and only single chops are opened into the echo.
Organ bubble: a drawbar organ from Operator, and the first thing to go
Quick start
- Load
- Operator, Chorus-Ensemble, Amp, Cabinet, Channel EQ
- Set
- Operator four sine operators side by side; ratios A 0.5, B 1, C 2, D 3, levels 0 / −3 / −9 / −15 dB
- Chorus-Ensemble Vibrato, Rate 6 Hz, Amount 25%
- Amp Clean, Gain 3; Cabinet 1x12
- Channel EQ 80 Hz HP on, High −4 dB
- Part busy 1/16 figure around the skank, removed first at bar 21
- You should hear
- A drawbar-style organ bubble filling around the skank, the densest part and the first to go.
- Live edition
- Suite
The bubble is the busy 1/16 organ figure that fills in around the skank. It is the most rhythmically dense part, so the version removes it first (bar 21) and brings it back only briefly. Live has no organ instrument, so Operator’s additive algorithm stands in for drawbars.
Piano: the intro line and the offbeat double
Quick start
- Load
- Electric, Channel EQ, Compressor, Saturator
- Set
- Piano Electric piano preset, bounced
- Channel EQ 80 Hz HP on, Low −3 dB, High +1 dB
- Compressor 3:1, attack 10 ms, 3 dB; Saturator Analog Clip 2 dB
- Send B Spring −12 dB at rest, 0 dB on bar 4
- Parts a dry intro line, then the skank doubled an octave up, quietly
- You should hear
- A short dry line announcing the tune, then a quiet hammered edge on the skank, thrown to the spring.
- Live edition
- Standard and up
The piano has two jobs. In the intro it plays a short dry line that announces the tune, which is the only time it is heard alone. In bars 15–20 and 57–64 it doubles the skank an octave up, quietly, so the chop has a hammered edge. It is the channel most often thrown to the spring rather than the echo.
Horns: a two-bar fanfare, a four-bar line, and the last note thrown
Quick start
- Load
- Channel EQ, Compressor, Saturator
- Set
- Horn stem own recording or Wavetable brass, bounced
- Channel EQ 80 Hz HP on, Mid +2 dB 1.5 kHz
- Compressor 4:1, attack 5 ms, 4 dB; Saturator Analog Clip 3 dB
- Send A Echo 0 dB on each phrase’s last note, channel muted on the next beat
- Send B Spring −15 dB
- You should hear
- A fanfare and a four-bar line whose last notes fly off into the echo as the channel cuts.
- Live edition
- Intro and up
The horn section (trumpet and tenor sax in unison is the usual roots pairing, by ear) plays a two-bar fanfare at the top and a four-bar line in bars 41–44. The fader is open for the line, and the last note of each phrase is sent to the echo at full level while the channel is muted on the next beat. For stabs built from synths, see Horn stabs. A line works better played than programmed: Strings, brass & physical models has the stock routes.
Melodica: four bars in the second half, then an echo of itself
Quick start
- Load
- Operator, Auto Filter
- Set
- Operator melodica patch, on channel 7 with the horns, never together
- Auto Filter HP 24, 300 Hz
- Send A Echo Pre, 0 dB on held notes
- Send B Spring −10 dB
- Part enters at bar 45 for four bars; bars 63–64 only as echoes of one note
- You should hear
- Four bars of melodica, then only its echo, returning as repeats of a single note.
- Live edition
- Suite
Build the patch from Dub siren and melodica. In this track it shares channel 7 with the horns and never plays at the same time as them. It enters at bar 45, plays four bars, and returns at bars 63–64 only as echo repeats of one note sent pre-fader.
The vocal side: one line, one word, and a fader that stays down
Quick start
- Load
- Channel EQ, Auto Filter, Compressor
- Set
- Vocal stem the full A-side take on channel 8, fader at −inf
- Channel EQ 80 Hz HP on, Mid +2 dB 2 kHz
- Auto Filter BP 1.2 kHz, Res 20%; opened to HP 150 Hz for the dry line
- Compressor 4:1, 5 dB
- Sends Echo and Spring pre-fader, −inf at rest, opened for a single word
- You should hear
- One dry line in bars 21–24, then the voice heard only as single words in the echo.
- Live edition
- Intro and up
A version keeps a fragment of the A-side vocal. Here the whole vocal is on channel 8 for all 72 bars with its fader at −inf. In bars 21–24 the fader opens for one line. After that the vocal is heard only through the echo, from a pre-fader send opened for a single word. For slicing and replaying words from pads, see Vocal fragments. This version deliberately does not cut the vocal: it uses the timing the singer already gave it.
The tape echo: a three-head Echo rack with a head selector
Quick start
- Load
- Audio Effect Rack, Echo, EQ Eight
- Set
- Head 1 Echo 1/8, FB 45%, chain 0 dB
- Head 2 Echo 1/4, FB 45%, chain −2 dB
- Head 3 Echo 3/8, FB 45%, chain −4 dB
- Each head Stereo, Link on, synced, Reverb 0, Duck 0; EQ Eight HP 280 Hz, LP 3 kHz
- Chain Selector zones as in the rack, Fade 0, on a Mode macro
- You should hear
- An echo with five voices: switching heads changes the repeat pattern, and a head joining mid-tail is part of the sound.
- Live edition
- Suite
The return itself is built in the Dub delay send rack (loop filter, Saturator, Limiter, Throw macro). This track swaps its single Echo for three “heads” that a Mode macro switches between, taking the idea from the RE-201’s documented three playback heads and rotary selector. The page assumes the heads are evenly spaced (1:2:3) at 1/8, 1/4 and 3/8. That is a simplification of a real transport, chosen so every head lands on the grid at 72 BPM.
| Mode (selector) | Heads on | Repeats after a hit at 72 BPM | Used for |
|---|---|---|---|
| 0–24 | 1 | 417, 833, 1250 ms… (1/8) | Rim slap, tight fills |
| 25–50 | 1 + 2 | 417, 833 (both), 1250, 1667… | Skank throws: busy, ends on the beat |
| 51–76 | 1 + 2 + 3 | 417, 833, 1250 (all stack up) | The drops: densest, runs away fastest |
| 77–102 | 2 + 3 | 833, 1250, 1667, 2500… | Vocal words: one beat of space first |
| 103–127 | 3 | 1250, 2500… (3/8) | Horn and melodica last notes, across the bar line |
- Each chain is an Echo in Stereo mode with Link on, synced, the delay time set per head, Reverb 0 and Duck 0. Each EQ Eight copies the loop band so the heads match.
- Set the Chain Selector zones as in the rack, Fade 0. The Mode macro steps through five combinations, and a head joining or leaving mid-tail is part of the sound.
- The Echo filter is in each loop, but the heads do not feed each other the way a single tape does. Pushing the shared Feedback macro gets close. For true cross-feeding, send Return A to itself, which is a specific kind of chaos (see the desk approach on Dub FX).
Feedback and filter rides on the echo across the 72 bars
Quick start
- Load
- Auto Filter
- Set
- Feedback 40% in the riddim, 55% under the voice, 85→100% in Drop 1, cut on the last beat of bar 31
- Head mode 1+2+3 in both drops, 2+3 under the voice
- Return LP 5 kHz, closing 5 k → 600 Hz in Drop 1; open 5 kHz in the HP drop
- Wobble 10%, rising to 35% in the outro
- You should hear
- The runaway sinks underwater in Drop 1, the echo is the only bright thing in the HP drop, and the outro tails sound like tape slowing down.
- Live edition
- Intro and up
The mechanics of runaway and return sweeps are covered in Feedback runaway and Filter sweeps on returns. Here is the score: where in this track the feedback climbs, where the head mode changes, and where the return filter closes.
Spring return: one crash per phrase, and the piano falling in
Quick start
- Load
- EQ Eight, Saturator, Hybrid Reverb, Utility, Limiter
- Set
- Sends rim −15, piano −12, horns −15 dB, Return A −12 dB
- EQ Eight HP 300 Hz (48 dB), LP 5 kHz (12 dB)
- Saturator Analog Clip, 6 dB
- Hybrid Reverb Convolution, spring IR, Decay 1.6 s, Vintage on; Utility Mono; Limiter −3 dB
- Crash send pushed to 0 dB for the accented snare ending each fill bar
- You should hear
- A mono spring crash at the end of every phrase, with the echo repeats splashing into it.
- Live edition
- Suite
Build the spring from Spring reverb, a mono convolution spring at about 1.6 s. In this track it plays a fixed role: a crash on the accented snare that ends every fill bar (bars 4, 12, 20, 28, 32, 40, 48, 52, 64), done with a send pushed to 0 dB for that hit alone, plus a low constant send from rim, piano and horns. Return A also feeds the spring at −12 dB (a return-to-return send), so the echo repeats splash into it the way the RE-201’s built-in spring sits after its heads.
The Big Knob: a ten-position stepped high-pass from a Chain Selector
Quick start
- Load
- Audio Effect Rack, EQ Eight
- Set
- Rack ten chains, one EQ Eight each with two 12 dB low-cut bands; chain 1 bypassed
- Positions 70 Hz, 100 Hz, 150 Hz, 250 Hz, 400 Hz, 700 Hz, 1.2 kHz, 2 kHz, 4 kHz
- Chain Selector zones 13 wide, every Fade 0
- Macro Chain Selector mapped to “Big Knob”, last on the Riddim group
- You should hear
- A high-pass that jumps in clicking steps instead of sweeping smoothly, thinning the riddim like a switched desk filter.
- Live edition
- Standard and up
A swept Auto Filter is smooth, while the documented Tubby filter was a switched passive unit that moved in steps and clicked. This rack copies the stepping with stock devices: ten chains, each holding one EQ Eight high-pass at a fixed frequency, and a Chain Selector mapped to one macro with no crossfade, so turning the knob jumps from one filter to the next. It sits on the Riddim group for the HP drop, and a copy sits on channel 7 for the horn squeal. The ten frequencies are the page’s own choice. They do not claim to match the Altec unit.
- Make an Audio Effect Rack with ten chains, an EQ Eight in each, and enable two low-cut bands (12 dB each) at the frequency of that position. Chain 1 is a bypassed EQ Eight.
- Open the Chain Selector, give each chain a zone 13 wide as in the rack, and set every Fade range to 0, so there is no crossfade and the steps click.
- Map the Chain Selector to a macro named Big Knob. Live’s automation stays continuous, but the zones turn any value into one of ten positions. That is the stepping.
- Put it last on the Riddim group, so the returns are not filtered, and a second copy on channel 7 for the horn squeal in bar 44.
Mutes as the composition: the channel map of the whole version
Quick start
- Load
- Utility, Auto Filter
- Set
- Drop-out one channel for 1–2 beats on a held Track Activator pad
- Strip all but the bass at once, or a macro muting each channel’s Utility
- Throw and mute send to 0 dB on the hit, mute on the next beat
- Ghost pre-fader send open with the fader at −inf
- Thin channel Auto Filter HP or the Big Knob instead of a mute
- You should hear
- Things leave on a beat, never on a downbeat, and the downbeat after each removal belongs to the bass.
- Live edition
- Intro and up
Seen from the desk, the arrangement diagram above is just a list of mutes. The grid below zooms into the busiest stretch of desk moves, bars 25–36, in beats, to show the grammar: things leave on a beat, never on a downbeat, and the downbeat after a removal belongs to the bass.
| Gesture | Where in this track | How in Live |
|---|---|---|
| Drop-out (one channel, 1–2 beats) | Hats, bars 27–28 and 44 | Track Activator on a pad, held |
| Strip (all but bass) | Bar 29 | Eight Activators at once: a Push 3 Mute press across selected tracks, or one macro on a rack that mutes each channel’s Utility |
| Throw and mute | Skank 41–44, horn 44, word 27 | Send to 0 dB on the hit, mute on the next beat |
| Ghost (dry closed, send open) | Vocal 27, melodica 63–64 | Pre-fader send with the fader at −inf |
| Thin (filter, not mute) | Bass 27–28, drums 33–36, riddim 49–52 | Channel Auto Filter HP, or the Big Knob on the group |
Recording the mix: automation passes into Arrangement, then one printed take
Quick start
- Load
- Auto Filter
- Set
- Controller Track Activators 1–8 on pads, sends on knobs, Big Knob, Mode, Feedback and Return A’s Auto Filter cutoff on encoders
- Passes Automation Arm and Arrangement Record: mutes, then sends, then filter and feedback, then faders
- Edit only mistakes, with Simplify Envelope; leave throws 30 ms late
- Print audio track on Resampling, loop bars 1–72, comp the best four-bar phrases
- Re-Enable Automation before every pass
- You should hear
- The version played as a performance, layered or printed, keeping human throws and the one perfect runaway.
- Live edition
- Intro and up
The track is a performance, so record it as one. There are two ways to do it, and the page uses both. Method A records mixer moves as automation in layered passes, the way you would overdub parts. Method B prints the whole mix to audio in looped full-length passes and comps the best bars from the take lanes, the nearest Live gets to a one-take desk mix.
- Map the controller first. Track Activators for channels 1–8 go on pads (momentary for drop-outs). Send A and Send B of channels 4, 6, 7 and 8 go on knobs. The Big Knob macro, the Three-Head Echo’s Mode and Feedback macros, and Return A’s Auto Filter cutoff go on encoders. Push 3’s Mix mode and track Mute buttons cover most of this without MIDI mapping.
- Pass 1, mutes only: turn on Automation Arm and Arrangement Record, play bars 1–72 and play only the mute pads. The manual documents that the Track Activator records as automation.
- Pass 2, sends: play again, touching only the send knobs. Pass 3: Big Knob, Mode, Feedback and the return filter. Pass 4: faders (the bass push, the outro fade).
- Edit only mistakes. Select a messy send lane and use Simplify Envelope, then nudge a late mute breakpoint onto the beat. Do not tidy the throws, because a throw that lands 30 ms late is a human throw.
- Method B, print: make an audio track with input Resampling, loop bars 1–72 and arm it. Record three passes while riding feedback and throws on top of the automation. Each loop pass becomes a take in its own take lane. Audition the takes and comp the best four-bar phrases into the main lane, cutting on phrase boundaries (bars 4, 8, 12…) so echo tails are not chopped.
The master bus: a two-track machine, not a mastering chain
Quick start
- Load
- Saturator, Glue Compressor, EQ Eight, Utility, Limiter
- Set
- Saturator Analog Clip, Drive 2 dB, Soft Clip on
- Glue Compressor 2:1, attack 30 ms, release Auto, 1–2 dB
- EQ Eight LP 15 kHz (12 dB), low cut 25 Hz (12 dB)
- Utility Bass Mono 120 Hz
- Limiter −1 dBTP, only on runaways
- You should hear
- A master that behaves like a two-track tape machine: a little warmth and glue, catching only the runaways.
- Live edition
- Standard and up
The version was printed to a two-track tape, so the master is treated as that machine: a small amount of tape-style saturation, a gentle glue, a high cut, and a limiter that only catches the runaways. For tape character in more depth, see tape saturation and wow and flutter.
Mix decisions
Levels and headroom
| Source | Peak at rest | Peak at its loudest moment | Note |
|---|---|---|---|
| Bass | −10 dBFS | −8 dBFS (Drop 1 push) | Loudest channel. Everything is set relative to it |
| Drums (kick + rim) | −13 dBFS | −11 dBFS (fill snare) | Kick about 3 dB under the bass on a meter |
| Skank, organ, piano | −20 / −24 / −22 dBFS | Same | Guests stay quiet. The echo makes them seem bigger |
| Horns / melodica | −18 dBFS | −16 dBFS | Held notes for the echo |
| Return A (echo) | −14 dBFS | −3 dBFS (Limiter ceiling, runaway) | The biggest peaks in the track come from here |
| Return B (spring) | −16 dBFS | −6 dBFS (crash) | Crash peaks are short. Let the master Limiter see them |
| Main | about −6 dBFS | −1 dBTP (runaway) | About −12 LUFS integrated, and the drops measure quieter |
EQ carving
Bus structure
Width per element
| Element | Width | How |
|---|---|---|
| Bass, kick, rim, skank, vocal | Mono, centre | Utility Mono on each stem |
| Hats, organ | Narrow stereo (about 60%) | Utility Width 60% |
| Piano, horns | Slightly off-centre (L15 / R15) | Pan, the way a four-channel desk spread its guests |
| Echo, spring | Mono, centre | Utility Mono after each return: the repeats sit behind the source, not around it |
Page recipe. A mono-return dub sounds smaller in headphones and much bigger on a sound system, because every repeat sums straight into the stack. For a headphone release, widen Return A to 120% with Utility and keep Bass Mono at 200 Hz, as the send rack does. For cutting a dubplate, see Mixing for sound systems.
Build it yourself in 10 steps
- 72 BPM, E minor, two chords (Em and D). Program a one-drop kit with 1/8-triplet hats, a fill every fourth bar, and the bass line, all for 72 bars.
- Add the guests for the full length: guitar skank, organ bubble (Operator additive), piano double, horn fanfare and line, melodica, and a sung vocal. Record what you can play.
- Bounce everything to eight audio stems (drums, hats, bass, guitar, organ, piano, horns/melodica, voice). From here on, no editing.
- Give each channel Channel EQ and an Auto Filter HP, group the eight as Riddim, and set the rest levels in the table, bass first.
- Return A: the Three-Head Echo rack (1/8, 1/4, 3/8 heads, loop band 280 Hz–3 kHz), Saturator, Limiter −3 dB, Mono. Return B: a mono spring with a driven Saturator. Send A to B at −12 dB.
- Build the Big Knob: ten EQ Eight chains on a Chain Selector with Fade 0, last on the Riddim group.
- Map a controller: mutes, sends, Big Knob, echo Mode and Feedback, return filter.
- Pass 1: record the mutes for the whole 72 bars, following the arrangement diagram. Pass 2: sends and throws.
- Pass 3: Big Knob, echo feedback and return filter, so Drop 1 at bar 29 is a mute drop and bar 49 is the stepped HP drop, with the return on bar 53. Pass 4: bass push and outro fade.
- Print three passes to a Resampling track, comp the best phrases, master gently (Analog Clip, Glue 2:1, LP 15 kHz, Limiter −1 dBTP), and check that the drops are still quieter than the return.
Variations
| Variation | What changes |
|---|---|
| Four-track discipline | Only four stems: drums+hats, bass, rhythm (guitar, organ and piano bounced together), voice/horns. No hat-only mute, so drop-outs become HP moves on the drum channel, and the guests can only leave together. Closer to the documented early-70s setup, and it forces bolder moves. |
| Rockers / steppers | Kick on every beat (rockers) or every 1/8 (steppers), tempo 74–78, head mode 1 most of the time. The return in bar 53 brings the bass in on its own for one bar before the kick (see the transition note). |
| Showcase (vocal + dub) | Put the vocal side in front, 64 bars with the full vocal, and run this version straight after it with no gap. The listener hears the song, then its dub, which is the order the 12-inch “discomix” of the later 70s used (by ear). |
| Melodica version | The melodica replaces the vocal as the voice: it plays the line in bars 21–28, and its fragments are thrown in the drops. Head mode 3 (3/8) on its held notes. |
| Digital dub (80s) | Drum-machine kit, keyboard bass, Delay in place of Echo for clean repeats, no Wobble, and a sharper Big Knob (48 dB EQ Eight bands). See the sub-styles table on the genre page. |
Live performance version
A dub version is already a live performance, so the live set is the same desk with two changes. The riddim runs from Session as long clips, so a scene can be extended when the room wants it. And the Arrangement automation is replaced by your hands, with the Session record button capturing the night.
Scenes
| Scene | Blueprint bars | Clips playing | Manual moves |
|---|---|---|---|
| 1 Intro | 1–4 | All eight channels, 4-bar clip | Mute all but horns and piano. Spring crash on bar 4 |
| 2 Riddim | 5–20 | All eight, 16-bar loop | Open drums and bass, then hats, then guitar and organ |
| 3 Voice | 21–28 | All eight, 8-bar loop | Vocal fader for one line, then pre-send words. Hat drop-outs |
| 4 Dub section | 29–48 | All eight, 4-bar loop (repeat as long as wanted) | The strip, feedback runaway, rim-only return, horn and melodica throws |
| 5 HP drop | 49–52 | All eight, 4-bar clip, no loop | Big Knob staircase, feedback to 100%, cut on beat 4 |
| 6 Return / outro | 53–72 | All eight, 16-bar loop | Big Knob off on the downbeat. Everything open, then strip to drum and bass |
Performance rack
Push and controller mapping
- Faders: an eight-fader MIDI controller on the eight channels. Push 3 Mix mode shows volume and sends on its encoders when no fader box is to hand.
- Mute pads: Track Activators on eight pads in momentary mode for drop-outs, plus one pad for the strip (all but bass).
- Throw knobs: send A on guitar, piano, horns and voice (pre-fader on voice and melodica). Rest at −inf, so nothing reaches the echo unless you put it there.
- Encoders: Big Knob on the encoder with the clearest detents, if your controller has them, since the steps should feel like steps. Then Echo Mode, Feedback, Tone.
- Capture: Session Record with Automation Arm on records your moves into the scene clips. Arrangement Record captures the whole set as a linear take you can edit into a release.
Hardware integration
- Set the Hardware Latency of each External Audio Effect once with a click. A tape echo returning 8 ms late is fine, because tape drifts anyway. A spring returning late smears the crash off the snare.
- A real tape echo’s feedback (its intensity control) can run away faster than Echo’s. Keep the Limiter after the return even with hardware.
- If the stems go out to an analog mixer, use its own high-pass on the channels as the Big Knob, and its mute buttons as the Track Activators. Live is then just the tape machine, which is the documented relationship.
- See Hardware and modular integration for routing, and reel echo for using a tape machine as a delay.
Related
- New alongside this page: Famous dub tracks, the real records behind this archetype
- Dub reggae, the genre page whose sounds this track places; Dub techno, UK bass 140, Deep dubstep, Trip-hop and downtempo
- Dub FX & delays, Sub bass, Chords and stabs, Keys and mallets, Guitar and band instruments, Strings, brass and physical models, Voice and vocals
- Dub delay send rack, Dub chord rack, Performance rack, Feedback rack
- Delay design, Reverb and space, Automation and modulation, Tape, lo-fi and analog character, Mixing for sound systems, Performance and live, Resampling workflow
- Racks and macros, Operator, Routing and resampling, Hardware and modular integration
- Dub techno track, the same desk logic at 120 BPM; Jungle track, where the dub sub went next; Deep dubstep track
- BPM to ms, Note to Hz, Genre tempo map