ABLETON 12 SOUND DESIGN ATLAS
Home › Track Deconstructions

Famous Deep & Dub Techno Tracks — How They Were Made

Twelve landmark deep, dub and hypnotic techno records, from Basic Channel to Voices From The Lake. For each: what is documented about how it was made, what is by-ear analysis, and how to rebuild the techniques in Live 12, with the filter movement as the main thread.

dub technodeep technominimalbasic channelmauriziorhythm and soundmonolakeplastikmanjeff millsrobert hoodgasporter ricksechospacedeepchordvoices from the lakefilterdelaychain reaction

This canon runs on very few notes. A Basic Channel record can hold one chord for twelve minutes, and a Gas track can loop the same few seconds of orchestra for an hour of album. What moves is the filter on the chord, on the delay feedback and on the whole mix, plus the decision to perform the mixdown by hand instead of programming it. This page covers twelve records that set the rules, in roughly chronological order. Each analysis separates Documented facts (linked to the interview, lecture or review they come from) from By ear analysis (listening, or the common understanding among producers). These artists are notoriously guarded. Basic Channel in particular are, in Attack Magazine’s words, “notoriously tight-lipped” about gear. So the “By ear” labels are deliberate, and most of what circulates online about their setups is folklore. The Live 12 recreations rebuild the technique, not the record. Every piano-roll example is a generic voicing that demonstrates the idea, not a transcription.

Documented
Stated by the artist, a collaborator or a primary-source feature (RBMA lecture/feature, Sound On Sound, RA, The Quietus, Frieze, label/artist site). Link given inline.
By ear
Our listening analysis or long-standing community consensus. Treat it as a hypothesis to test against your own copy.
BPM / key
Unless linked, these are by-ear estimates. Drop the record into Live, let warping detect the tempo, and check the key with a tuner on the lowest sustained tone. Vinyl pitch and later remasters both shift these.
Assumed tools
Live 12.2+ Suite (Auto Filter redesign, Roar, Echo, Hybrid Reverb, Drift, Meld). Everything also works with Live 12.0 and stock devices, except where a 12.2 feature is named.

Overview of the twelve tracks

Track / recordYearArtistSignature technique
Phylyps Trak1993Basic ChannelFast 909/conga engine under hiss-coated stabs, with the mix performed by hand on faders
Quadrant Dub1994Basic ChannelA vocal house track “versioned” into long chord-and-bass dub. Filter and echo throws are the arrangement
M41995MaurizioChord stab and delay merged into one rolling texture, with the low-pass slowly breathing over dozens of bars
Mango Drive1998Rhythm & SoundA reggae riddim re-versioned “completely live in the mix”: muted channels, spring/tape echo, skank chords
The Bells1994/1996Jeff MillsOne percussive bell motif plus a TR-909 that is “the backbone”. Hypnosis through repetition and live drum-machine performance
Minimal Nation (Rhythm of Vision)1994Robert HoodSubtraction: 909 + Juno 2 “grey area sound” + pocket sequencer, no effects on the recording
Cyan1995–97MonolakeSelf-made drums, a slowed-down mosquito as the bass, field recordings, and an unsynced Juno arpeggio
Biokinetics1996Porter Ricks“Underwater” techno: rolled-off highs, panned gurgles, the 4/4 kick as a door into sound design
Königsforst1998–99GasClassical samples processed by “zoom, loop and alienation” into a fogged loop over a muffled kick
Consumed1998PlastikmanOdd-length step sequences driving resonant filters. “An album of feedback”, recorded live in one take
The Coldest Season2007Deepchord pres. EchospaceAll-analogue chain: Space Echo and Echoplex tape delays plus noise generators. The delay is the instrument
Voices From The Lake2012Donato Dozzy & NeelHypnotic, linear builds with deep foreground/middle/background layering, edited down from a month of studio jams

The tracks

Basic Channel: Phylyps Trak (1993): the hiss-coated engine

1993 · Basic Channel BC-02 · ≈130–134 BPM (by ear) · key: minor-ish, single-chord (by ear) · Level: intermediate

Why it matters. This is the second Basic Channel 12", and it shows the dub-techno vocabulary before it slowed down: a driving, busy 909 track with stabs, granular-sounding chords and hiss on top. The B-side “Phylyps Base” shows the other half of the method, which is to strip the same material back into a dub. Kirk Degiorgio’s RA “Rewind” review calls it “a stripped-down rework, less busy and perhaps more effective for it”.

Documented. Release data (BC-02, 1993) and the later “Phylyps Trak II” (BC-09, 1994) are on Wikipedia’s Basic Channel discography. The RA review describes a fast tempo, a 909 drum part, a conga rhythm, stabs and tape-hiss-like noise, with tape delay and reverb in the texture. It also notes that Basic Channel records were pressed in Detroit. Moritz von Oswald’s RBMA lecture confirms it: “All the Basic Channel and Maurizio records were cut in Detroit.” The Quietus’ 25-year retrospective says it is built around a triplet-figure bassline. An earlier Quietus feature credits part of the sound to “von Oswald’s insistence on doing all mixing manually, hands on faders”.

By ear / folklore. The chord synth is widely assumed to be a Sequential Prophet-5, but that is community consensus, as Attack says explicitly, not a confirmed fact. RA mentions a rumoured Waldorf Wave. The hiss reads as a constant bed under everything rather than a property of any one channel.

Signature moves in Live 12:

  1. Triplet-figure bass under a straight kick. Write the bass in 1/8T against a four-on-the-floor kick. The phase relationship keeps shifting through the bar, so a static two-note line feels like it moves. Use Drift with Osc 1 Sine plus a touch of Osc 2 Saw, the MS2 low-pass at about 300 Hz and a short filter envelope.
  2. Hiss as glue, sent to the delays as well. Put a looped noise or vinyl bed on its own track at about −30 dBFS, with a high-pass at 1 kHz and a low-pass at 9 kHz. Also give it a small send to the chord delay, so the echoes carry hiss and blend into the bed.
  3. Hands-on-faders mixdown. Map chord level, delay send and conga level to a controller and record the arrangement as one pass, as automation. Do not draw it.
  4. The “Base” version. Duplicate the arrangement, mute the conga and half the hats, open the delay feedback, and pull the chord low-pass down to 600–900 Hz. That is your dub version.
What to steal: make two versions of every track, a full “Trak” and a stripped “Base”, from the same parts. The stripped one is often the better record.

Basic Channel: Quadrant Dub (1994): versioning into a long dub

1994 · Basic Channel BC-06 · ≈124–128 BPM (by ear) · single sustained minor chord colour (by ear) · Level: intermediate

Why it matters. This is the template for the long-form dub techno track: a chord, a bass and a kick, with all development carried by echo throws and filtering. Attack Magazine uses “Quadrant Dub 1” as a reference track in its dub chord tutorial.

Documented. BC-06, 1994 (Wikipedia). The Quietus writes that Quadrant Dub reworks the duo’s own vocal house track “I’m Your Brother” into nearly forty minutes of chord-heavy, bass-driven material. The source material came first and the dub is a version of it, which matches von Oswald’s own description of the dub process in the RBMA lecture: “Usually, it works from a vocal track to an instrumental.” He also says automation was effectively impossible at the time, so the movement was played on the desk.

By ear. The chord stab is short and its echo is the sustained element. The perceived chord brightness changes over long spans, consistent with a low-pass being ridden by hand. Low end stays mono and steady while everything above about 300 Hz drifts.

Signature moves in Live 12:

  1. Start from a “vocal” version. Build a fuller arrangement first (chord, bass, a vocal or lead line, busier drums), then make the dub by muting rather than writing. Group everything and add channel mutes as Session clips, or use Arrangement lanes.
  2. Stab-plus-echo chord. Play a short stab (amp decay 120–200 ms) into Echo set to dotted 1/8 left and 1/4 right, feedback 55–70%, with Echo’s own filter set to a high-pass at about 250 Hz and a low-pass at about 3 kHz. The dry stab sits about 4 dB under the echoes.
  3. Echo throws. Use a second send with Echo at 100% wet and 85% feedback, whose send level rests at −inf. Push it up to −6 dB on single stabs (one beat), then pull it back so the tail rings out.
  4. The 64-bar filter ride. Automate the chord’s Auto Filter (LP 24 dB, resonance 25–35%) from about 700 Hz up to 3.5 kHz over 32 bars, then back over 32. Record it with a knob, not with a mouse. Small wobbles are the point.
What to steal: write the “song” first and dub it second. Subtraction from something real gives a better dub than a minimal sketch does.

Maurizio: M4 (1995): chord and delay as one breathing texture

1995 · Maurizio M-4 · ≈125–130 BPM (by ear) · one minor chord colour (by ear) · Level: advanced

Why it matters. The Maurizio M-series is the purest form of the idea. The chord is not a part on top of the groove, it is the groove, smeared by delay into a continuous rolling pulse, and the low-pass opens and closes over spans so long that you only notice in retrospect.

Documented. Release data: M-4 (1995) and M-4.5 (1995), with the M-Series compilation in 1997 (Wikipedia). Cut in Detroit and import-only, per von Oswald (RBMA). RA’s Phylyps review notes NSC dead-wax stamps on Maurizio releases indicating Detroit mastering and pressing (RA). Von Oswald describes the general process as “mainly constantly running” and the mixing as live. No track-specific gear for M4 is documented.

By ear. The stab repeats at high density (8th or 16th rhythm) and the delay is set so the repeats land between stabs, so there is no clear dry/wet boundary. The cutoff movement is slow and asymmetric: it opens over many bars and closes faster. There is noticeable resonance at the cutoff, giving a vowel-like “wah” only when it moves.

Signature moves in Live 12:

  1. Dense stab pattern + interleaving delay. Stabs on an 8th-note figure with gaps; Delay (not Echo, for cleaner repeats) in Sync mode at 1/16 L and 3/16 R, feedback 60%, Filter on with BP center around 1.2 kHz, width 6–8 (narrower = more “dub”).
  2. Resonant low-pass on the sum of stab + delay. Group chord and delay on one Audio Effect Rack, then Auto Filter (12.2) in the Clean or OSR circuit, LP 24, resonance 35–45%, cutoff 500 Hz → 2.5 kHz.
  3. Asymmetric 64-bar automation. Open over 48 bars, close over 16 — draw it once, then use clip envelope in a long (64-bar) unlinked loop so it runs independently of a 1- or 2-bar chord clip.
  4. LFO micro-breath. Add an LFO device (Live 12 stock) at 0.07–0.1 Hz, depth 8–12%, mapped to the same cutoff — it rides on top of the long automation so nothing is ever static.
What to steal: filter AFTER the delay, not before. Filtering the source only darkens the attack, while filtering the sum makes the whole echo-cloud breathe.

Rhythm & Sound: Mango Drive (1998): live-in-the-mix riddim versioning

1998 · Rhythm & Sound RS-02 (paired with “Mango Walk”) · tempo: reggae one-drop feel, warp to check · minor (by ear) · Level: intermediate

Why it matters. This is the point where the Basic Channel duo goes fully into Jamaican dub practice: a riddim, a version and a mix performed in real time. It explains where dub techno’s chord-and-echo behaviour came from.

Documented. Released in 1998 as RS-02 alongside “Mango Walk” by the Chosen Brothers (Wikipedia). In his RBMA lecture von Oswald says it directly: “this track was done completely live in the mix.” He also says the “guitar-sounding synth chords” carry over from the source, that “the bass is modified and simplified into the new track”, that he favours dub-style mixdown straight to stereo tape, and that he likes old analogue tape echoes “which are still available as digital copies”. The duo’s work with Lloyd “Bullwackie” Barnes in New York is covered in The Quietus.

By ear. The chord “skank” sits on the off-beat with a very short envelope and dark tone. Bass is round and simplified. Whole instruments drop out for bars at a time (channel mutes), and echo tails continue through the mutes.

Signature moves in Live 12:

  1. Mute-performance setup. Give every part its own track, map the track activators to Push or a controller, and arm Arrangement record. Perform 6–8 minutes of mutes and unmutes in one take, the way von Oswald describes.
  2. Echo throws that survive mutes. Put the echo on a return with pre-fader sends (set the send to Pre in the mixer). Muting the track stops the dry sound but not the echo, because the throw has already been sent. This is the main dub mechanic.
  3. Simplified bass. Reduce your riddim bass to root and fifth, mostly half notes. Use Operator with a sine on A, B at ratio 1 and level −30 dB for a hint of grit, then a low-pass at 250 Hz.
  4. Tape-echo character. Echo with its Character section’s Noise and Wobble turned up (about 15–25%), feedback 70%, with Echo’s filter set to a high-pass at 400 Hz and a low-pass at 2.5 kHz. Every repeat gets darker and thinner, as a worn tape echo does.
What to steal: pre-fader sends plus mute performance. Most of the dub feeling comes from echoes that outlive their source.

Jeff Mills: The Bells (made 1994, released 1996): one motif, a performed 909

Made 1994, released 1996 · Purpose Maker · ≈135–140 BPM (by ear) · tonal centre from the bell motif (by ear) · Level: intermediate

Why it matters. It is not dub techno, but it is the hypnotic-techno archetype: one percussive melodic idea repeated until it becomes a trance state, and drums that are played rather than programmed. Every deep/hypnotic DJ track that relies on a single motif owes something to it.

Documented. According to XLR8R, Mills made it in 1994 and played it from a custom 13" record before releasing it officially on Purpose Maker in spring 1996. It was later performed with the Montpellier Philharmonic (2005). Gearnews’ Jeff Mills feature says the track builds from a single percussive idea, with the TR-909 as “the actual backbone of the track”, and describes Mills using the 909 as a live instrument, shifting beats, tweaking parameters and changing tempo. The synth used for the bell sound is not documented in these sources.

By ear. The motif is a short, bright, metallic figure with a fast decay. Its intervals stay fixed while the drums around it build and drop. The filter interest here is the 909 itself: hat decay, tuning and accent automation do the job that a chord filter does in dub techno.

Signature moves in Live 12:

  1. FM bell motif. Operator, algorithm B→A. A sine, B sine with Coarse ratio 3.5 (inharmonic), B envelope decay 300 ms, B level −12 dB. Amp decay 250 ms, sustain 0. Add Chorus-Ensemble at a low amount for shimmer. Keep the motif 1–2 bars long and loop it without variation.
  2. Motif brightness ride. Instead of a filter, automate Operator’s Osc B level (the FM index) over 32–64 bars. Brightness then rises as more harmonics appear, not just as louder top end.
  3. 909 played live. Drum Rack with 909 samples. Map hat decay, the clap/snare chain activator and the kick tune to Push encoders. Record 64 bars of hand-tweaking as automation. Use Note Echo or Beat Repeat (Grid 1/16, Chance 30%) on the hat chain for Mills-style stutters.
  4. Tempo push. Mills is described as changing tempo live. Automate the song tempo +1–3 BPM over a long section for tension. Echo and Delay in Sync mode follow along.
What to steal: lock the melodic material and perform the drum machine. Variation lives in the drums, while the motif is the ritual.

Robert Hood: Minimal Nation (1994): subtraction as composition

1994 · Axis · tracks in the ≈130–140 BPM range (by ear) · focus track “Rhythm of Vision” · Level: beginner–intermediate

Why it matters. This record is the source of the word “minimal” in techno, and the main counter-model to dub techno’s fog: no reverb wash, few parts, all tension from looping and dropping.

Documented. RBMA’s track-by-track feature lists the gear: a Roland TR-909 (bought from Gerald Donald of Drexciya), SH-101, Juno 2, Juno 106, a Yamaha DX100 borrowed from Jeff Mills, a Yamaha QY pocket sequencer and a four-channel mixer. There were no effects during recording, and mastering engineer Ron Murphy added reverb “here and there”. It was composed in fall 1993 and finished in winter 1994 on secondhand and pawn-shop equipment. “Rhythm of Vision” is built on a Juno 2 sound Hood called a “grey area sound”. The feature sums up the concept as “a new sound forged not through addition, but subtraction.” “One Touch” was punched in live on the QY for human feel.

By ear. The chord/riff is a short, dry, mid-range tone repeated on an off-beat grid. Changes come from muting and unmuting and from the 909’s hat and ride patterns. The “filter” is mostly the synth’s cutoff knob turned slightly during the take.

Signature moves in Live 12:

  1. Grey-area stab. Drift: Osc 1 Rectangle, Osc 2 Saw −12 st, MS2 LP at 1.4 kHz, resonance 20%, filter envelope amount +20, decay 180 ms, amp decay 220 ms. Mono voice mode. Nothing after it except Utility.
  2. Four-channel constraint. Limit yourself to four tracks (kick, hats, rhythm synth, riff synth), Hood’s mixer count. Every new idea must replace an old one.
  3. Manual cutoff in the take. Map the Drift filter frequency to a knob and record ±15% movement over a 64-bar take, small enough that it reads as tension, not as a sweep.
  4. Mastering-only reverb. If you want space at all, use one short Hybrid Reverb send (Quartz, decay 0.8 s, 10%) on only one element for a few bars, as a mastering-stage touch.
What to steal: the four-track limit and “no effects until mastering”. That forces arrangement by subtraction, which is the core skill in hypnotic techno.

Monolake: Cyan (1995–97): self-made drums and the pre-Ableton lab

Recorded 1995–97 · Chain Reaction (album Hongkong, CRD-04, 1997) · ≈120–128 BPM (by ear) · Level: advanced

Why it matters. Robert Henke and Gerhard Behles went on to co-found Ableton, and their studio software from this period was the testbed for Live. Cyan was the first Monolake track, and it shows a different route into dub techno: field recordings and synthesis experiments in place of a Prophet and a 909.

Documented. From Henke’s RBMA interview: Cyan was the first Monolake track. They had no classic drum machines, partly for lack of money and partly by preference, so the drums are a “strange synthesis” of their experiments. The bassline is a heavily slowed-down mosquito recording. Birds and crickets from a forest recording appear. Behles played an arpeggio on an unsynced Juno 6 and matched its tempo by hand. Henke credits the result to “technical shortcomings and complete inexperience”. Henke’s own release page dates Hongkong (CRD-04, Chain Reaction, 1997) and lists Cyan at 12:04, with material created 1995–97 that includes field recordings from Hong Kong and Guangzhou. CDM covers their 1995 PX18 step sequencer/timeline, “the spiritual and technical predecessor to Ableton Live”, which was built around variation and rotation.

By ear. The bass has a buzzing, slightly unstable pitch, which fits a slowed organic recording. The arpeggio drifts against the grid. The track’s depth comes from an ambience bed that never stops, plus a darkened, filtered delay.

Signature moves in Live 12:

  1. Organic bass from a field recording. Load a short insect, buzz or motor recording into Simpler (Classic mode), warp off, and transpose −24 to −36 st, so slowing it down turns it into a low drone. Use Auto Filter LP 24 at about 180 Hz, resonance 30%. Play roots via MIDI. The pitch wobble is a feature.
  2. Self-synthesised drums. Kick from Operator (sine, pitch envelope +24 st, 40 ms) or Drum Sampler’s synth modes. Hats from Operator noise with a short envelope through a narrow band-pass at 8 kHz. Do not use 909 samples.
  3. Hand-synced arp. Record an arpeggio from a free-running source (Eurorack, TD-3 internal clock, or Arpeggiator with Sync off and Rate in Hz). Then warp it with only a few warp markers so it drifts a little against the grid between markers.
  4. Ambience bed + filtered delay. Run the forest/city recording on a looped track at −28 dB. Add Delay at 3/16, feedback 65%, with the band-pass filter swept by a slow LFO between 600 Hz and 2.4 kHz.
  5. PX18-style rotation. Use the Live 12 MIDI Transformation Recombine (rotate) or Time Warp on the arp clip every 16 bars. That is variation by rotation, not by writing new notes.
What to steal: constraints and found sound. Henke’s “technical shortcomings” turned into a signature, so let an unsynced, slightly wrong element stay in your track.

Porter Ricks: Biokinetics (1996): techno heard from underwater

1996 · Chain Reaction (CD), reissued Type TYPE100 2012 · ≈125–130 BPM (by ear) · focus track “Biokinetics 1” · Level: advanced

Why it matters. It took the Chain Reaction aesthetic to its most abstract: a submerged 4/4 pulse as the frame for sound design and stereo imaging. It is the root of much “deep” and “hypnotic” techno sound design since.

Documented. Andy Mellwig is a mastering and cutting engineer at Berlin’s Dubplates & Mastering, and Thomas Köner has a background in film sound engineering (RA review). That review describes “Biokinetics 1” as loops arriving within about 30 seconds, a gurgling sound and a frantic metallic sound panned against each other, and percussive “splat” noises on each bar. Köner explains the concept in Time Out: sound underwater loses its high frequencies, so the music “isn’t sharp”. He also gives the frame: “Porter Ricks used a valid formula—four to the floor—and used it as a key to open a door.” Crack Magazine notes that the aquatic quality was not intended, and that the process starts from “a very centric, core element and process” and grows as the system runs. Mellwig writes the duo’s own software. No 1996 hardware list is documented.

By ear. The kick is muffled, with almost no click above about 2 kHz. The gurgles are band-pass-filtered, resonant, modulated noise or feedback. The stereo field is wide but the low end is narrow. Highs are rolled off on the bus, not per element.

Signature moves in Live 12:

  1. Underwater bus filter. On the master or a group: Auto Filter LP 12 at 3–5 kHz, resonance 10%. Automate the cutoff between 2 and 8 kHz over 64+ bars, so the listener “surfaces” and “dives”.
  2. Panned gurgle pair. Two tracks of noise from Operator (Noise waveform) or Meld (Noise Flow). Each gets Auto Filter BP with resonance 60–70%, cutoff modulated by its own LFO (S&H on one, Sine 0.3 Hz on the other), panned L35 and R35, with opposite LFO phases so they argue across the stereo field.
  3. Splat on the bar. One-hit short noise burst through Resonators (Mode A, decay short) into Grain Delay (spray 40 ms), triggered once per bar.
  4. Self-growing system. Feed the gurgle’s return into Roar in Feedback routing, feedback low, with Roar’s filter as a band-pass. Ride feedback amount with a macro. A small core process grows into the track, as Köner describes.
What to steal: filter the mix, not just the parts. A slow low-pass on the whole bus is the strongest “depth” control in this genre.

Gas: Königsforst (1998–99): orchestral fog, zoomed and looped

Late 1990s (album dates differ between sources: 1998/1999) · Mille Plateaux · ≈110–125 BPM, some tracks beatless (by ear) · Level: intermediate

Why it matters. Wolfgang Voigt showed that the dub techno “chord” does not have to come from a synth. A few seconds of late-Romantic orchestra, looped and smeared, can do the same job, with the kick as a heartbeat under a forest of sound.

Documented. Simon Reynolds in Frieze: Gas was a “lab project”. The music is built from small samples of classical records, mostly late-19th-century (Wagner, Berg, Schoenberg), processed by “zoom, loop and alienation”, with vinyl hiss audible. Königsforst is a forest near Cologne, and Voigt described his teenage visits as “Hansel and Gretel on acid”. Rolling Stone quotes him calling Gas “sample-based music”, notes that sources include “old Arnold Schoenberg symphonies and Richard Wagner operas”, and that the 1990s sound was deliberately rough: he didn’t care about “proper sound”. The four late-90s albums run up to Pop (2000).

By ear. The loops are short (fractions of a bar to a couple of bars), low-passed and heavily reverbed, so the orchestral source is unrecognisable but keeps its string/brass timbre. The kick is soft and muffled, and the loop seams are smoothed by crossfades and reverb.

Signature moves in Live 12:

  1. Zoom. Take 1–3 seconds of an orchestral recording you have the rights to use (or your own string/brass patch from Live’s orchestral packs, resampled). Put it in Simpler in Classic mode with Loop on, length 1/2 to 1 bar, Snap off and a crossfade of 30–60%.
  2. Loop. Warp in Texture or Complex Pro mode, and tempo-sync the loop to a multiple of the bar so it cycles hypnotically against the kick.
  3. Alienation. Auto Filter LP 24 at 1.2–2 kHz, then Hybrid Reverb (Dark Hall, decay 6–9 s, 35% wet), then Vinyl Distortion (crackle low) or a hiss layer, then Utility at 120% width.
  4. Fog filter ride. Automate the low-pass between 700 Hz and 3 kHz over 64–128 bars. Add a second, slower ride on Hybrid Reverb wet (25 → 50%) so the loop sinks deeper as it brightens.
  5. Heartbeat kick. Soft kick: Operator sine, short pitch envelope, amp decay 350 ms, then Auto Filter LP 12 at 400 Hz to kill the click. A distant thud, not a punch.
Sampling commercial classical recordings has rights implications even if the composition is public domain. Record or render your own source for release work.
What to steal: any harmonically rich source can be your “dub chord”. Low-pass it, loop it and reverb it until it is an atmosphere.

Plastikman: Consumed (1998): odd-length sequences driving resonance

1998 · Minus / Novamute · slow, ≈110–120 BPM range (by ear) · title track focus · Level: advanced

Why it matters. It is the darkest, most minimal end of the hypnotic spectrum, and the clearest documented example of filter movement as composition: step sequencers of unusual lengths animating resonant filters, everything cross-modulated, recorded live.

Documented. Sound On Sound’s Classic Tracks feature gives the gear: a Doepfer MAQ16/3 sequencer (one original plus three newer units) sending 16-step patterns to a Serge modular, a Roland TR-909 used mostly as a sequencer, a Kawai XD-5 drum synth for tom/conga sounds, an Akai S3000 for some kicks, an Atari ST looping effects automation, an Allen & Heath mixer with MIDI mute automation, and DAT recording edited in Sound Forge. On the title track, an XD-5 tom/conga sound is opened and closed through a Serge filter with heavy effects. An 8- or 16-bar Atari loop switched the effects mutes at set points. The machines ran overnight and the take was recorded live with minor edits. About ten minutes in, a Serge riff distorts from excessive resonance and was kept. The MAQ ran cyclical five- and twelve-note sequences driving the Serge filters. The 909 shuffle sat around 3, and triplet delays added syncopation. Quotes: “Consumed is an album of feedback. Everything was cross-modulating everything else.” “Everything was recorded live and in one take.”

By ear. The low end is a dark, sub-heavy pulse. The “melody” is mostly resonance peaks moving, and the space comes from long reverbs being switched in and out rather than faded.

Signature moves in Live 12:

  1. Polymetric filter sequencing. Make three clips that drive cutoff, each with a different loop length: 5 steps, 12 steps and 16 steps of 1/16. In Live 12 the simplest route is unlinked clip envelopes: on one MIDI clip, unlink the envelopes and give the Auto Filter cutoff a 5/16 loop, the resonance a 12/16 loop, and a macro offset a 1-bar loop. For true per-step control, use three Shaper devices (stock Max devices) mapped to the same cutoff, each with a stepped breakpoint shape and a different sync rate.
  2. Tom/conga into resonant LP. A tom from Drum Sampler or Operator (sine, pitch env +12 st, decay 180 ms) into Auto Filter (12.2) using the OSR or MS2 circuit, LP 24, resonance 70–85%, drive 6 dB. Let it nearly self-oscillate. If it cracks, keep it.
  3. Mute-switched effects. Create a long reverb return (Hybrid Reverb, decay 12 s) and a 1/8T delay return. Automate the send activators or the return track mute on/off in an 8- or 16-bar cycle, as hard switches rather than fades.
  4. Cross-modulation. Envelope Follower on the kick mapped to the tom filter cutoff (inverted, 20%), and a second Envelope Follower on the tom mapped to reverb send. Everything modulates everything.
  5. Overnight take. Record the master to a new audio track for 20–40 minutes while you only touch two or three things, then edit by cutting and reordering, as Hawtin did in Sound Forge.
What to steal: odd-length modulation loops on the cutoff. Five against twelve against sixteen gives evolving filter movement from three tiny patterns.

Deepchord presents Echospace: The Coldest Season (2007): tape delay as the instrument

2007 · Modern Love · ≈117–122 BPM (by ear) · focus track “Celestialis” · Level: intermediate

Why it matters. This is the second-generation classic that made dub techno into a whole weather system. Delays and noise are not effects added to chords. They are the main sound sources, with chords and kick as triggers.

Documented. Rod Modell (Detroit) and Steve Hitchell (Chicago). Gridface’s review says it was made with only analogue equipment, including a Roland Space Echo and Maestro Echoplex tape delays plus various noise generators. “First Point of Aries” opens with a wash of static and synths echoing over rain-like hiss. “Celestialis” is “very Basic Channel-like”, recalling “Octagon”. Release data is on Wikipedia. In The Quietus, Modell says he carries “small recorders everywhere” and fills tracks with processed field recordings, that tracks usually start as 30–40 minute pieces edited down to 6–10 minutes, and that he “never consider[s] the dancefloor when making a track.”

By ear. Chord stabs are almost buried. What you hear is their tape-echo regeneration, getting darker with each repeat. The noise bed is loud, sometimes rivalling the chord. The delay time is nudged during the take, causing pitch smears.

Signature moves in Live 12:

  1. Echo as a tape machine. Echo on a return, sync off, Left 370 ms and Right 490 ms (unrelated times), feedback 75–85%. Echo’s filter set to a high-pass at 300 Hz and a low-pass at 2 kHz, Noise on, Wobble 20–30%, and Reverb inside Echo at 20%.
  2. Time nudges = pitch smears. Map Echo’s delay time to a knob, and in the take nudge it ±10–20 ms for a beat or two. The repeats warble like a tape machine being pushed.
  3. Delay-feedback filtering. Map Echo’s low-pass to a macro and automate 1.2 → 3.5 kHz over 64 bars. Then ride feedback 70 → 90% in 16-bar waves and pull it back before runaway. Put a Limiter at −3 dB on the return as a safety net.
  4. Noise generators. Two noise sources: Operator Noise through Auto Filter BP at 5 kHz (wind), and a field-recorded rain loop with a high-pass at 800 Hz. Both at −26 to −30 dB, with a send to the echo as well.
  5. Long take, then edit. Record 30–40 minutes of performance and cut to 8.
What to steal: unsynced delay times plus a performed time knob. Synced dotted-8th delays sound neat, while free times and nudges sound like weather.

Voices From The Lake: Voices From The Lake (2012): layered hypnotic builds

2012 · Prologue (PRGCD001) · ≈120–128 BPM where there is a kick (by ear) · focus track “Iyo”/album arc · Level: advanced

Why it matters. This is the modern hypnotic-techno touchstone. Instead of fog, it has a very clear depth map: foreground, middle ground and background, and builds so gradual that the kick arriving feels like an event.

Documented. RA’s review: released February 2012 on Prologue (PRGCD001), recorded in Rome in summer 2011, and used as the basis for live sets. Most sounds have no real-world reference, and the sound design has distinct foreground, middleground and background. The album has an overture-like opening and works in groups of three or four tracks toward peaks. Neel had mastered much of Dozzy’s earlier work. Dekmantel’s feature says it began as a live set for Labyrinth festival in Japan. Neel says they “recorded material for more than ten albums before putting all the elements into place”, and Dozzy says they were in the studio together for over a month. Modular systems are mentioned in passing. In an Arturia artist story on his later setup, Dozzy describes his approach: arpeggios and revolving patterns (Riley/Reich), minimal arrangements, elliptical drum patterns, gradually shifting atmospheres, and shaping a bass-lead gradually with the master cutoff.

By ear. The drums are soft and clicky, almost liquid. Sequences are short and polymetric. Each layer has its own filter trajectory, so the mix keeps reorganising its depth without changing volume much.

Signature moves in Live 12:

  1. Three-plane depth map. Foreground: dry, band-limited (high-pass 400 Hz) clicks and sequence, Utility width 40%. Middle: arp through Auto Filter BP plus a short Echo. Background: pad/noise through Hybrid Reverb at 100% wet with a 1.5 kHz low-pass, and width 130%.
  2. Revolving arpeggio. Use the Arpeggiate transformation or the Arpeggiator device (Up/Down, 1/16, steps 2 oct) on a held 3-note chord. Insert Shape or Seed to regenerate the source chord every 32 bars, so it stays the same pattern but slowly changes colour.
  3. Per-layer filter trajectories over 128 bars. Foreground high-pass slowly falls (600 → 250 Hz). Middle band-pass rises (800 Hz → 2.2 kHz). Background low-pass slowly opens (1 → 4 kHz). Each one moves in a different direction, and the depth shifts.
  4. Master-cutoff bass-lead. A Drift or Meld bass-lead in mono with a single long low-pass ride, 200 Hz → 1.6 kHz over 64 bars, as described in Dozzy’s later method.
  5. Jam then curate. Record long multitrack jams, keep perhaps 10%, and sequence tracks in groups of three or four toward a peak.
What to steal: give each depth plane its own filter that moves in a different direction. The mix then stays in motion without anything getting louder.

Patterns across the canon

Across 1993–2012 the same few rules recur. Most are documented by the artists themselves: live mixdowns (von Oswald, Hawtin), long takes edited down (Hawtin, Modell, Dozzy & Neel), restraint and subtraction (Hood), found or sampled sources instead of preset synths (Henke, Voigt, Modell). The rest are by-ear constants: hiss beds, dark kick tops, chords whose echoes are louder than the chord itself.

Filter movement summary

MoveWhere it shows upLive 12 implementationTime scale
Resonant LP on the chord+delay sumMaurizio, Basic Channel (by ear)Auto Filter LP24 after Echo/Delay inside one rack; Res 30–45%32–64 bars per open/close, asymmetric
Delay-feedback filteringEchospace (documented tape delays), Rhythm & SoundEcho’s own HP/LP inside the feedback loop; LP 1.2 → 3.5 kHz ride; feedback waves 70 → 90%16-bar feedback waves inside 64-bar tone ride
Hand-ridden mixdown filterBasic Channel “hands on faders” (documented)Map cutoff to a hardware knob, record automation in one passWhole track
Odd-length sequenced cutoffPlastikman (documented 5- and 12-step MAQ sequences)Unlinked clip envelopes 5/16, 12/16, 1 bar on one cutoff macroCycles realign every 3–15 bars
Bus low-pass “depth”Porter Ricks (documented high-frequency loss concept)Auto Filter LP12 on group/master, 2–8 kHz64–128 bars
Fog LP + reverb rideGas (by ear)LP24 on loop + Hybrid Reverb wet automated in parallel64–128 bars
Opposing per-layer filtersVoices From The Lake (by ear; documented depth layering)HP falls, BP rises, LP opens on three planes128 bars
Timbral “filter” via FM index or synth cutoffJeff Mills, Robert Hood (by ear)Operator Osc B level / Drift cutoff ±15% in the take8–32 bars, subtle
Common practice: never let two big filter moves peak at the same bar. Offset the chord-sum ride and the delay-tone ride by 8–16 bars, so there is always one thing opening while another settles. That is how a one-chord record stays interesting for ten minutes.

Related