ABLETON 12 SOUND DESIGN ATLAS

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
Labels on this page: Documented means a linked source supports the practice. By ear is listening and scene knowledge, not fact. Everything else is the page’s own recipe: a Live 12 way to get the result, not a claim about how any record was made. No real record is transcribed. The piano roll and step grids are generic shapes. The Famous dub tracks page covers specific records.

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:

Make the riddim yourself, or use one you have permission to version. A dub of somebody else’s multitrack is still their recording. Every part on this page can be played or programmed in Live.

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.

The return works because of the one drop: the bass has beat 1 to itself and the kit arrives on beat 3. A steppers kit (kick on every beat) would cover the bass’s return, so on a steppers version, bring the kick back one bar later.

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

Instrument: Arrangement audio tracks · Level: essential · Genres: dub, roots reggae, version

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.ContentsSendsFader at rest
1 DrumsKick, rim, snare, toms (one stereo stem)Spring only (rim, snare)−13 dBFS peak
2 HatsClosed and open hat, shakerNone−22 dBFS peak
3 BassBass (mono)Never−10 dBFS peak
4 GuitarSkank chopEcho (throws), spring−20 dBFS peak
5 OrganBubbleEcho (rare)−24 dBFS peak
6 PianoOffbeat double + intro lineEcho, spring−22 dBFS peak
7 Horns / melodicaFanfare, horn line, melodica (shared channel, never together)Echo, spring−18 dBFS peak
8 VoiceThe vocal side’s lead, uncutEcho (pre-fader), spring−inf (opened for fragments)
  1. 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.
  2. 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.
  3. 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.
  4. Group the eight channels as “Riddim”. Returns A and B and the master bus sit outside the group.
Restricting yourself to four channels (drums, bass, rhythm, voice) is the strictest version of this discipline. See four-track bounce and the four-track variation below. You lose the hat-only mute, but the drum channel’s high-pass becomes a hat solo.

One-drop drums on two channels: the kit, the shuffle hats and the fill into each phrase

Instrument: Drum Rack (Drum Sampler pads) · Level: beginner · Genres: roots reggae, dub, one drop

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.

  1. 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.
  2. Rim Track Delay +8 ms, kick 0. The late rim against a quantised bass is the drag the genre page describes, set once here.
  3. 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

Instrument: Analog or recorded bass DI · Level: essential · Genres: dub, roots reggae

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.

Do not send the bass to the echo “just for the drop”. A bass repeat at 625 ms lands on the next bass note and the floor turns to mud. If the drop wants a bass echo, send a copy high-passed at 300 Hz, as Dub FX warns.

Guitar skank: in for 14 bars, then only ever heard as a throw

Instrument: Guitar DI → Amp → Cabinet, or Electric · Level: beginner · Genres: roots reggae, dub

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

Instrument: Operator · Level: intermediate · Genres: roots reggae, dub, rocksteady

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

Instrument: Grand Piano (Suite Pack) or Electric · Level: beginner · Genres: roots reggae, dub

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

Instrument: Recorded horns or Wavetable · Level: intermediate · Genres: roots reggae, dub

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

Instrument: Operator (reed patch) · Level: intermediate · Genres: dub, roots, rockers

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

Instrument: Audio (the vocal stem) · Level: intermediate · Genres: dub, version

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

Device: Audio Effect Rack (Echo × 3, Chain Selector) · Level: advanced · Genres: dub, roots, dub techno

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 onRepeats after a hit at 72 BPMUsed for
0–241417, 833, 1250 ms… (1/8)Rim slap, tight fills
25–501 + 2417, 833 (both), 1250, 1667…Skank throws: busy, ends on the beat
51–761 + 2 + 3417, 833, 1250 (all stack up)The drops: densest, runs away fastest
77–1022 + 3833, 1250, 1667, 2500…Vocal words: one beat of space first
103–12731250, 2500… (3/8)Horn and melodica last notes, across the bar line
  1. 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.
  2. 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.
  3. 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).
The single-Echo 3/16 setting from the genre pages is still the best all-round dub time. The three-head rack is here because a version uses the echo as an instrument, and changing heads gives that instrument five voices.

Feedback and filter rides on the echo across the 72 bars

Device: Three-Head Echo macros, Auto Filter on Return A · Level: intermediate · Genres: dub

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

Device: Hybrid Reverb (Convolution, spring IR), Corpus · Level: beginner · Genres: dub, roots

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

Device: Audio Effect Rack (10 × EQ Eight, Chain Selector) · Level: advanced · Genres: dub, roots, dub techno

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.

  1. 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.
  2. 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.
  3. 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.
  4. 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.
Some chain switches will click, especially at low positions on the bass. That is part of the effect, but a click on a bass note peak can be 6 dB above the music. Check the group meter at position 2 and 3. If a click overloads, raise those two chains’ HP by 10 Hz rather than adding a fade, because a fade turns the steps into a sweep.

Mutes as the composition: the channel map of the whole version

Device: Track Activators, Push 3 or a MIDI controller · Level: essential · Genres: dub, 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.

GestureWhere in this trackHow in Live
Drop-out (one channel, 1–2 beats)Hats, bars 27–28 and 44Track Activator on a pad, held
Strip (all but bass)Bar 29Eight 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 muteSkank 41–44, horn 44, word 27Send to 0 dB on the hit, mute on the next beat
Ghost (dry closed, send open)Vocal 27, melodica 63–64Pre-fader send with the fader at −inf
Thin (filter, not mute)Bass 27–28, drums 33–36, riddim 49–52Channel Auto Filter HP, or the Big Knob on the group
Five gestures and no more. The desk approach lists the same vocabulary. This table is where each one goes in this particular version.

Recording the mix: automation passes into Arrangement, then one printed take

Workflow: Automation Arm, Arrangement Record, Resampling, take lanes · Level: advanced · Genres: dub, any live-mixed music

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.

  1. 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.
  2. 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.
  3. 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).
  4. 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.
  5. 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.
If you touch an automated control between passes without recording, Live overrides its automation and lights the Re-Enable Automation button. A version where the vocal send is quietly overridden plays as if the throw never happened. Press Re-Enable Automation before every pass.
Why comp at all, if the documented method was one take? Because the comp is made of whole takes, it is still a set of performances, and it lets you keep the one runaway that came out perfectly. If you want the strict version, print one pass and keep it.

The master bus: a two-track machine, not a mastering chain

Device: Saturator, Glue Compressor, EQ Eight, Limiter · Level: beginner · Genres: dub, roots

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

SourcePeak at restPeak at its loudest momentNote
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 dBFSSameGuests stay quiet. The echo makes them seem bigger
Horns / melodica−18 dBFS−16 dBFSHeld 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
Mainabout −6 dBFS−1 dBTP (runaway)About −12 LUFS integrated, and the drops measure quieter

EQ carving

Bus structure

Width per element

ElementWidthHow
Bass, kick, rim, skank, vocalMono, centreUtility Mono on each stem
Hats, organNarrow stereo (about 60%)Utility Width 60%
Piano, hornsSlightly off-centre (L15 / R15)Pan, the way a four-channel desk spread its guests
Echo, springMono, centreUtility 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

  1. 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.
  2. 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.
  3. Bounce everything to eight audio stems (drums, hats, bass, guitar, organ, piano, horns/melodica, voice). From here on, no editing.
  4. 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.
  5. 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.
  6. Build the Big Knob: ten EQ Eight chains on a Chain Selector with Fade 0, last on the Riddim group.
  7. Map a controller: mutes, sends, Big Knob, echo Mode and Feedback, return filter.
  8. Pass 1: record the mutes for the whole 72 bars, following the arrangement diagram. Pass 2: sends and throws.
  9. 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.
  10. 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

VariationWhat changes
Four-track disciplineOnly 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 / steppersKick 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 versionThe 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

SceneBlueprint barsClips playingManual moves
1 Intro1–4All eight channels, 4-bar clipMute all but horns and piano. Spring crash on bar 4
2 Riddim5–20All eight, 16-bar loopOpen drums and bass, then hats, then guitar and organ
3 Voice21–28All eight, 8-bar loopVocal fader for one line, then pre-send words. Hat drop-outs
4 Dub section29–48All eight, 4-bar loop (repeat as long as wanted)The strip, feedback runaway, rim-only return, horn and melodica throws
5 HP drop49–52All eight, 4-bar clip, no loopBig Knob staircase, feedback to 100%, cut on beat 4
6 Return / outro53–72All eight, 16-bar loopBig Knob off on the downbeat. Everything open, then strip to drum and bass

Performance rack

Push and controller mapping

Hardware integration

Related