Minimal Techno in Bitwig: a course that ends with a finished track

Eight sessions of 60 or 90 minutes, roughly four weeks at two sessions a week. Every session ends with a saved result that cannot be reworked in the next one. The goal is not “learning production” but closing one track of about six minutes and putting it out.

How the course is built

In minimal techno almost everything is decided at the level of groove and time, not at the level of how many sounds there are. In a track with four elements every element is heard in full, so the cost of a mistake in timbre or timing is high, while the cost of a mistake in “composition” is close to zero: there is no composition. Hence the order: first a rhythmic foundation you can listen to for three minutes without boredom, then bass, then exactly one characteristic element, then movement over time, and only then arrangement and mix.

Mixing sits at the end not on principle but because balancing a loop and balancing a track are different jobs. In a loop everything plays at once and you want to make everything loud. In a track half the elements exist only in specific sections, and what felt overloaded in the loop turns out to be empty in the breakdown.

The main risk here is not technique, it is an unfinished project. So three rules are built in:

1. A timer per session. When time is up the session is over and the file is saved with a new version number. Decision quality in production rises for the first forty minutes or so, then plateaus and starts to fall: your hearing tires and edits become random.
2. No new plugins or sample packs after Module 0. Browsing libraries feels like work but produces no decisions. A limited toolset forces you to process the sound you have instead of hunting for a new one, and processing is the genre.
3. Decisions get committed. Every module ends with a “commit” step, usually Bounce in Place or saving a preset. An uncommitted decision stays an open question and pulls attention through every following session.

How to read this course

Each module has a “What to do in Bitwig” section, which is the list of actions, and everything else. Everything else is written for the case where an action does not give the result you want and it is unclear what to turn. While the result is fine, skip the explanations: they answer questions you do not have yet.

Ear training

Three distinctions, one per session, each takes about twenty minutes to install and then works forever. Without them minimal feels like shooting blind, because decisions there are made right at the edge of noticeability.

1. Kick body. Make two kicks with different amounts of energy around 60 to 90 Hz and switch between them at medium volume. Listen not for “which is louder” but for “which is heavier”.
2. A 10 ms shift. Make two hat clips, one exactly on the grid, one shifted by 10 ms, and switch. At first you hear no difference, then you hear it always.
3. Slow modulation. Play a loop with a 32 bar modulator and without it, and learn to notice that something is changing without looking at the screen.

Project constraints

Write these down before starting and do not change them:

• Tempo: 128 BPM (the genre runs roughly 125 to 132)
• Key: A minor or F# minor, set as Project Key Signature (Bitwig 6, key K snaps notes to the key)
• Maximum 8 sounding tracks, buses not counted
• Length: 6:00 to 6:45
• One synth, one sampler, everything else built in effects
• Release deadline: a specific date four weeks out

On key: minimal is rarely melodic, but you still need a key, because the kick, the bass and the resonances of percussion all have pitch. If they disagree the track sounds muddy, and that is fixed by transposing, not by EQ. Minor is chosen for practical reasons rather than aesthetic ones: in minor it is easier to build mood neutral chords with a seventh that pull the track neither toward sadness nor toward joy.

On 8 tracks: this is not a limit of taste but a limit of attention. Eight elements fit in your head at once and you can remember what each of them does. At twelve you start managing a list instead of working with sound.

Module 0. References and template

Goal: understand by ear how time works in the genre, and build a project where routine will not distract you later.

Why references matter specifically here

In genres with melody and vocals a reference is for sound. In minimal a reference is for time: how long a state lasts, how many bars until the next event, how small an event can be and still count as an event. This is the one thing tutorials cannot give you, because tutorials always show a loop, never six minutes.

How to listen

1. Pick 3 references from the list in the appendix.
2. Listen with a stopwatch and fill a table: time, what came in, what went out, what changed without coming in or going out (filter, delay, level). The third column is the important one: you will find that most events in minimal are not “a new sound appeared” but “one parameter of an existing sound moved”.
3. Count how many elements sound at once at the densest point. Usually four to six, which is surprising.

Twenty minutes of this beats any article about arrangement, because the table is yours, with real timings, and in Module 6 you can build your own map from it.

The template: what and why

The track list is not project decoration, it is a hypothesis about what the track consists of, declared in advance. While the tracks are empty, adding a ninth is psychologically easy. Once the roles are declared and named, adding a ninth becomes a visible violation rather than invisible drift. It works like a cast list in a play.

Roles are separated along two axes at once: frequency and event density. Two elements coexist fine if they differ on at least one axis, and start fighting if they match on both.

Then the utility tracks:

• Drums group (Kick, Hats, Perc, Ghost). The group exists not for tidiness but for two specific moves. First: one filter across the whole group for breakdowns, where the entire rhythm rides up and comes back. Doing that with four separate automation lanes is both painful and worse sounding, because the elements move out of sync. Second: light shared compression that glues the drums into one layer instead of three independent ones.
• Effect Track with a short reverb (0.6 to 1.2 s) and an Effect Track with a long one (3 s and up). Two sends because the short reverb says “these sounds are in one room” while the long one says “there is space here”. With one you have to choose between glue and depth.
• Effect Track with a delay, time 3/16 or dotted 1/8. In minimal a delay is a separate rhythmic instrument rather than an effect, so it gets its own channel with a filter in the feedback path.
• Master with a Peak Limiter at the end. The limiter goes in from the start not for loudness but as a ceiling and an indicator. Set it so it barely engages, one to two dB on peaks. The point is to avoid discovering in Module 8 that a track mixed without a limiter changes character once one is switched on. The opposite risk exists too: mix deep into a limiter and the mix will be lifeless without any way to notice. So the rule is simple: the limiter is there but running almost idle until the last module.

About Bitwig editions

The device set depends on the edition: Essentials has 40, Producer 92, full Studio 154. Whole browser categories simply do not appear when there is nothing to show.

Check whether a device exists by searching its name in the browser, not by scrolling categories. Possibly missing from what this course mentions: Spectrum (the analyzer, Analysis category), Saturator, Humanize, some modulators.

A locked device looks like this: it loads, but instead of an interface you get a padlock. Clicking the padlock shows which edition you would need. Check at the start of a session, not when you hit the wall.

None of them is required. The course gives a substitute for each, and the analyzer is replaced by tuning by ear.

Other one time setup

• Set the tempo and the Project Key Signature.
• Put the references on the Reference track and match their level with a Tool device so it sits at your future mix level. Comparing against a louder reference is useless: louder always seems better.
• Save as a Template so the same track set is the starting point for the next tracks. Half the value of a template is that the second track does not start from nothing.

Done when: the template exists, you have a structure table for three tracks, and the project constraints are written down.

Materials

• Bitwig Learn: the official base of videos and articles
• A step by step breakdown of a Yuri Urano track in Bitwig with a downloadable project file. Worth opening someone else’s project just to see how few tracks are actually in it
• MusicRadar: 7 minimal production tips

Module 1. Kick and grid

Goal: one kick that holds the track, and an eight bar loop of kick and silence that is pleasant to listen to.

Why the kick comes first and alone

In minimal the kick occupies more of the mix than in any other genre, simply because there is almost nothing next to it. In a dense track a kick can be mediocre and other elements will cover it. Here it plays naked for several bars at the start and the end of the track, and the listener has time to examine it.

Second: the kick sets the low end budget. Everything decided about bass in Module 3 depends on how many hertz and how many milliseconds the kick took. So it cannot be chosen after the bass.

What a kick is made of

You do not need to read this in advance. It is for when the kick is done, something about it bothers you, and it is unclear what to turn.

A kick is not a “hit” but a short sound with falling pitch. For the first few milliseconds it is very high, then it drops fast and settles on one note. That sharp fall is exactly what the ear reads as attack. Hence three phases:

1. Click, the first 2 to 10 ms, the top of the fall.
2. Body, 60 to 90 Hz, lasting 60 to 150 ms.
3. Tail, a steady note at 40 to 60 Hz, decaying over 250 to 450 ms.

The practical value of the split is that three different complaints are fixed with three different knobs:

Short rule: a problem at the top of the spectrum is found at the start of the sound, a weight problem in the middle, a bass problem at the end.

What to do in Bitwig

1. Option A, fast: E-Kick, turn Decay, Pitch Amount, Click. Fine if you want to reach Module 2 today.
2. Option B, your own: Poly Grid, sine oscillator, AD envelope on amplitude, a second AD on pitch with a time of 20 to 40 ms and a depth of about 2 octaves, then Saturator. The pitch envelope is the click and the body: a fast pitch drop is perceived as attack. The kick also needs its very bottom trimmed, but that is hygiene and happens in Module 7.
3. Tune the pitch (see below). No more than ten minutes; this gets settled in Module 3 once there is a bass.
4. Make 4 versions, hear each in an 8 bar context, pick one, delete the rest. Delete, literally: saved alternatives bring the question back next session.
5. Bounce in Place. After that the kick is audio. This is not only about CPU: an audio file cannot be quietly tweaked, and it stops being a topic of discussion.

Tuning the kick

The tail of the kick holds one frequency for 300 ms, and the ear reads that as a note. So your kick has a pitch even if you never set one: it happened by accident. If that note disagrees with the key, kick and bass will sound muddy against each other, and EQ will not help.

The fastest method, with no extra devices at all:

1. Put a long A note on the bass track (for A minor) and play it together with the kick.
2. Turn the kick’s Pitch knob. If the pitches disagree you will hear beating: the level pulses in waves.
3. Stop where the pulsing slows down and disappears.

Frequencies if you want to check: E1 is 41.2 Hz, F#1 is 46.2 Hz, A1 is 55.0 Hz, C2 is 65.4 Hz.

The analyzer method exists but is less convenient than it sounds: the kick’s pitch falls, so instead of a clear peak you get a smeared blob. If you do want to look, the device in the browser is called Spectrum, in the Analysis category. Essentials and Producer may not have it, their device sets are cut down. This does not affect the result, the beating method is more accurate anyway.

How much this matters: more the longer the tail. With a 150 ms decay nobody will notice a semitone off.

Why the kick sits flat on four beats

The kick is the one element everything else is counted against. A percussion syncopation reads as syncopation only because there is something to push against. If the kick itself wanders, the reference point disappears, and with it the effect of everything you do in Module 2. The result is not more complex, just less legible.

Flat on four does not mean mechanically identical. Inside a straight grid there is plenty to move:

• Level: kicks two and four 1 dB below one and three.
• Timbre: slightly different tail length per hit, via a smoothed modulator.
• Omission: drop one kick every 8 or 16 bars.
• Shift the whole track 3 ms early or late. Changes the sense of rushing or leaning back across the entire track.

Dub techno, Romanian minimal, broken techno and electro all live happily without a straight grid, so this is not a law of the genre but a choice for the first track. A motionless foundation is what lets you hear what exactly you changed in the percussion. When everything moves at once you cannot tell what caused the effect. Break it on the second track.

Exercise. An 8 bar loop, kick only, listen for 3 minutes without touching the mouse.

Note what is actually being tested. This is not a test of “is it interesting” but of “is it irritating”. A single kick cannot be interesting, it has nothing to change. The question to yourself is whether you want to switch it off. If you do, find the complaint in the table above and fix it.

Done when: the kick is bounced, named, sitting on its track, and no alternatives remain.

Module 2. Hats, percussion, groove

Goal: a rhythmic loop that lives on its own.

The core: why an 8 bar loop holds for 3 minutes

This is the main content of the genre; everything else in the module serves this point.

The arithmetic is simple. The ear learns a repeating pattern in about three passes. Eight bars at 128 BPM is 15 seconds, so after 45 seconds the loop is fully known and there is nothing left to listen to.

Which inverts the task. Three minutes cannot be filled with an interesting loop. They can only be filled with something that is not learnable in 45 seconds. Inside those eight bars there has to be something with a period longer than eight bars, or something with no period at all.

The question is not “how do I make the loop richer” but “how do I make eight bars not be eight bars”. There are four ways.

1. Not one loop. If percussion runs at a length of 3 bars while everything else is 4, full alignment happens once every 12 bars. Add a third element 7 beats long and the period goes past a minute. You programmed nothing, but it can no longer be learned in 45 seconds. Costs one mouse move, so start here.

2. No two hits are identical. Two neighbouring hats differ in level, timbre, stereo position, micro shift. The ear compares each hit with the previous one, and while they differ the comparison continues. Once the hits are identical there is nothing to compare and attention leaves. This is exactly why velocity spread is done with a device rather than the mouse: a drawn spread is itself a pattern and gets learned along with the loop.

3. The pattern allows two readings. If an element sits so that it is unclear whether it is late for its own beat or early for the next, the ear cannot settle on an interpretation and keeps looking. This is what Villalobos and the Romanian scene are valued for: their loops literally cannot be listened to until they feel resolved. Practically: move a percussion element across sixteenths and listen not for “pretty or not” but for “am I sure where the one is”. The position where it gets slightly uncomfortable and you want to recount is usually the right one.

4. There is a hole in the loop. A place where the ear expects a hit and does not get one. An unclosed expectation works longer than a closed one: while a loop sounds complete you can put it aside, while something is missing you cannot. Practically: remove the most predictable hit, usually on beat four.

What this does not give you: none of the four makes a loop beautiful. They make it not boring, which is a different property. Beauty comes from the timbre of the characteristic element, that is from Module 4, and it has no business being in eight bars of drums.

Diagnosing by your own attention

Listen to yourself, not to the loop. Start it, note the time, mark the moment your thoughts wandered. Then use the table:

An honest caveat: three minutes of motionless listening on headphones is a harsher test than actual use. On a dancefloor the body picks up what attention does not. If the loop holds for two minutes, that is enough.

Why groove is micro timing, not pattern

A 4/4 grid at 128 BPM gives a sixteenth of 117 ms. The ear resolves shifts of 3 to 5 ms, so a single sixteenth contains dozens of distinguishable positions. Everything called groove in minimal happens inside that gap, not in the choice of which beat to put a hit on. Practical conclusion: keep the pattern primitive and work with shift and velocity.

Three sources whose period is longer than the loop

Methods 1 and 2 above are about pattern. But the most reliable sources are the ones with no period at all or a period longer than eight bars. There are exactly three, and all of them are needed here rather than in Module 5:

1. Continuous randomness. Randomize on velocity, Random on timbre. No period whatsoever, repetition is impossible in principle.
2. Slow modulation. An LFO of 8 or 16 bars on the percussion filter. The period is longer than the loop, so no two passes are the same.
3. A delay at a non integer time. 3/16 unfolds one hit into a 3 bar pattern.

Everything else, including any work on the pattern, lives inside a single bar and gets learned in half a minute.

The odd loop length trick and when it does not work

A clip of odd length (3 bars or 7 beats instead of 4 bars) shifts against the background. But it has a hard condition, without which it gives nothing at all:

the background must distinguish bars.

If the kick is flat on four and the hats are identical in every bar, the background structure is one bar long, not four. Against a one bar background a three bar clip is just a three bar clip: it repeats as it repeated, nothing shifts anywhere. The trick is useless here, and no clip length changes that.

The condition is met when the background has a phrase of 4 or 8 bars: the kick drops out every fourth bar, the hats have a different pattern every fourth bar, there is an accent on the first beat of the phrase. Then the odd clip lands in that phrase differently each time and the difference is audible.

Practical conclusion: do not use this on the first assembly of the loop. It belongs to the stage where a phrase already exists, that is the end of Module 2 or Module 6. Keep all clips at 4 bars until phrase structure appears.

And a limit for later: only make one or two elements odd. If all of them are odd the anchor disappears and the track stops moving.

What to do in Bitwig

1. Closed hat on eighths between the kicks, or on sixteenths. Do the velocity spread with a Randomize device in the Note FX chain, not with the mouse, and keep the depth small. The reason is simple: by hand you produce a pattern that repeats every cycle and becomes obvious within a minute. Randomize gives a new value every time, so repetition never occurs.

2. Micro timing: Humanize, or manual nudging of specific notes, or Groove in the project panel with Shuffle at 8 to 20 percent. Groove is global, which makes it the fastest way to hear whether swing is wanted at all.

3. One noisy or metallic sample in Sampler, short Decay, a filter after it, a slow LFO on cutoff with a period of 8 or 16 bars and small depth. Plus Velocity to the same cutoff, so the level spread from Randomize becomes a timbre spread as well.

4. Ghost percussion. Gives a long non repeating pattern out of a single note, because 3/16 does not fit a whole number of times into a bar: the repeats land in new places every bar and return to the starting position after 3 bars. In detail:

   Sound. Something indistinct with no pronounced attack: a quiet noise fragment, a reverb tail, a piece of field recording. Anything bright and identifiable (rim, clap, ringing metal) does not work, it demands attention.

   Sampler. Length 80 to 150 ms, set by the playback end field in the Offsets column with PLAY selected. A fade of 20 to 30 ms at the end so the cut does not click.

   Notes. One hit per bar, not on beat one: that is taken by the kick, and the repeats would land in the same places. Put it on the second or third sixteenth. Clip 4 bars, four identical hits inside.

   Delay+ after Sampler, not in its FX slot. Inside the slot the effect applies per note and the tails get cut off by new notes.

   Delay+ settings. Pattern: Stereo with a small Width; avoid ping pong, it draws attention. Time: base length 1/16, multiplier 3, offset neutral in the center. Or the same result via base 1/8 with a dotted offset, but not both at once. Feedback 40 percent, by ear four to six distinguishable repeats. Mix almost fully wet, the original hit is not needed. The FB FX button enables the filters that sit inside the feedback loop. Pull the high cut down: inside the loop rather than after it, each repeat passes through the filter again and gets duller than the last, and that reads as receding rather than as echo.

   Level. Mute the track and unmute it. Correct when muting makes things noticeably poorer but you cannot name what disappeared. If you can name it, turn it down.

   If it does not work. Heard as a separate part: too loud or too bright. Mush by the end of the bar: Feedback above 60, drop it to 30. Nothing changed: too quiet, or Mix is on dry.

5. Phrase: make something small differ every 4 or 8 bars. A dropped kick, a different hat pattern, an accent on beat one. Only after this does touching clip lengths make sense.

Three tracks besides the kick, no more: Hats (the straight grid), Perc (the groove itself), Ghost (background). Point 3 is a way to make a sound; it applies to any of them and is not a separate track.

Exercise. Assemble 8 bars of kick and percussion and listen for 3 minutes. If it is boring, first check whether the loop has even one source with a period longer than the loop. If it does and it is still boring, remove an element rather than add one: boredom in minimal often comes from masking, where two elements occupy the same zone and smear each other.

Done when: the loop of kick, hats and one percussion holds attention for 2 minutes and contains no more than three rhythmic sources.

Materials

• Polarity: building a techno track from scratch in Bitwig
• Loopmasters: 16 techno tips

Module 3. Bass

Goal: bass and kick as one organism rather than two competitors.

Why they cannot be made separately

Kick and bass occupy the same octave and sound at the same time. When two low frequency waves add up the result depends on their phase: some combinations reinforce, others partially cancel, and by ear that feels like “the kick disappeared” even though the level is unchanged. EQ does not fix it, because the problem is in time rather than in spectrum.

Two working solutions follow, and both are about time rather than frequency:

1. Separation by frequency. The kick gets the bottom, the bass gets everything above. This is done with cuts and belongs to Module 7.
2. Separation in time. The bass ducks for exactly the length of the kick hit. That is sidechain, and its original purpose was technical rather than stylistic: free up room, not create “pumping”. It happens here, because both the bass pattern and its envelope settings depend on it.

Why sidechain via a modulator rather than a compressor

A compressor reacts to the signal and has its own inertia: attack, release, threshold, knee. You cannot set the shape of the dip directly, only approximate it. The Audio Sidechain modulator in Bitwig is a follower envelope you can assign to any parameter at any depth and either polarity. The practical difference: you can duck not only level but also, say, the bass filter cutoff, and you can assign dips of different depth to different parameters at once. Plus a modulator adds no dynamics artefacts of its own.

The whole chain

Polymer (with its internal LP as timbre) then Distortion then Tool with the sidechain.

1. The synth

Polymer, monophonic mode. Mono because at low frequencies stereo is not perceived as width, and on a club system it can cause the channels to cancel.

Voice count is set not in the device but in the Inspector panel on the left, with the device selected. There are two monophonic modes: in one the voice is continuous and the envelope does not restart on a new note, in the other it does. For bass you want the second, every note needs its own attack.

The legato button next to Glide (it looks like a small padlock with an L) enables glide only for overlapping notes. Turn it on and set a small Glide: the bass stays crisp by default, and a slide appears wherever you deliberately extend a note so it runs into the next. This is classic 303 behaviour.

Amplitude envelope: A at zero, D short, S low, R short. Every note should be an impulse rather than a sustained tone.

2. Harmonics for small speakers

A phone speaker physically cannot reproduce 60 Hz. But the ear identifies pitch partly from the spacing of harmonics: if 120, 180 and 240 Hz are present, the brain reconstructs the missing 60. The phenomenon is called the missing fundamental. A pure sine has no harmonics at all, which is why a sine bass vanishes completely on a phone.

Three approaches, simple to complex:

Waveform. Not sine but triangle or saw, then close the LP. Zero extra devices, works in any edition.

Distortion. Distortion or Saturator after Polymer. Key controls: amount of distortion, symmetry (centered gives odd harmonics and a harder sound, offset gives even harmonics and a softer one), Mix at 30 to 50 percent so the clean low end survives. Turn the output down so switching the device on and off does not change the level, otherwise your judgement will be wrong. Only ever set it together with the kick; soloed, you always want more dirt.

A second layer an octave up via Instrument Layer, 12 to 18 dB below the main one, with an HP at 150 Hz. More control, but watch that the layer does not start sounding like a separate part.

Whether there are enough harmonics gets verified in Module 7 along with the other checks. For now the rule of thumb is simple: a pure sine definitely fails, a harmonic rich waveform or distortion definitely passes.

3. The filter

LP around 400 Hz inside Polymer. This is timbre; the filter envelope acts on it and every note gets its own brightness shape. A little resonance is fine and adds character.

General rule about resonance: a filter standing still needs no resonance. A filter that moves needs it, otherwise the movement is inaudible.

Cutting the bottom off the bass is hygiene rather than timbre, and it happens in Module 7.

4. The pattern

The bass goes between the kicks, not with them. Kick on the beats, bass on the “and” after each beat. The low end then sounds continuous while the two sources never coincide and physically cannot interfere.

Start like this: one note, four per bar on the eighths between kicks, short length, a 2 bar clip. Then three edits:

• Remove one note out of four, usually the last in the bar. A hole appears.
• Make one note a different pitch. A fifth down (7 semitones) is the most stable, an octave (12) gives movement without changing harmony, a minor third up (3) sounds darker.
• Make the two bars different: four notes in the first, three in the second.

Two or three pitches is enough: in minimal the bass is a rhythmic instrument with pitch, not a melody.

If the kick has double hits. A one off accent every 8 or 16 bars: just skip the bass note there, between two close hits it would not have time to speak anyway. If double kicks are part of the main pattern: the bass fills only the long gaps and stays silent in the dense places. Total low end density should stay roughly constant regardless of how much kick there is.

5. Sidechain

First judge whether it is needed. If the bass sits strictly between kicks with short notes, the effect will be subtle. Some overlap does exist: the kick tail lasts 300 ms while an eighth at 128 BPM is 234 ms, so the kick is still sounding when the bass note starts.

Steps:

1. Tool last in the bass chain.
2. In Tool’s modulator panel press plus and pick Audio Sidechain.
3. Choose the kick track in the input selector.
4. Start playback: the modulator listens to a real signal and shows nothing in silence.
5. Click the modulator’s name, it highlights.
6. Drag the Gain knob down. Depth must be negative so the level falls on the kick hit.
7. Click the modulator’s name again to leave assignment mode.
8. Set the recovery time so the bass is back by the next eighth.

Setting depth without numbers. By ear: raise from zero until the kick becomes clearer, as if it stepped forward. Keep raising until you hear the bass level pumping, which is already too much. Roll back to about halfway between those two points. As a magnitude, 6 to 10 dB, but the difference between 6 and 10 is barely audible so precision is unnecessary.

If Audio Sidechain is missing from your edition. A compressor with a sidechain input, source kick, 4 to 8 dB of gain reduction. Or no sidechain at all: shorten the bass release and the kick tail so they do not overlap. For minimal that is a working solution rather than a compromise.

If it does not work. Nothing happens: no input selected, the kick is not playing, or the depth was assigned upward. The bass disappears entirely: depth too large. Pumping appeared: recovery too slow. It got muddier: the problem was not overlap but frequency separation, and that is fixed in Module 7.

6. Check

Exercise. Mute everything except kick and bass. If the pair sounds like one continuous low frequency pulse, that is what you want. If you hear two separate parts, adjust the frequency separation or the sidechain recovery time.

Done when: the bass is bounced or its preset saved, and the pattern is no longer than 2 bars.

Module 4. One characteristic element

Goal: the thing the track will be recognized by. Exactly one of them.

Why exactly one

Kick, bass and hats are interchangeable in this genre: they perform a function and carry almost no individuality. Everything that distinguishes one track from another falls on one or two elements. With three or four of them they compete for attention and none is remembered. With one, the listener has no choice about where to look.

The second reason is practical: a single characteristic element is something you can afford to spend a long time on and make strange, because it is the only one.

Why the source should be your own

A library sample carries someone else’s decisions: their room, their processing, their attack character. Your own recording gives material nobody has heard, and that is the cheapest way to get unlikeness. Recording quality barely matters: in minimal the sound will be filtered, squashed and shortened beyond recognition anyway.

Stab and Texture: which one we make here

These are two roles from the template, and they differ in behaviour over time rather than in timbre.

Stab is a short sound with a clear attack that plays rarely: 200 to 800 ms, one event every 2 or 4 bars. It is noticed as an event.

Texture is a sustained sound with no pronounced attack that plays continuously. Only its disappearance is noticed.

This module makes the stab. Character comes from being noticed, and a sustained background creates no character. Texture is made in Module 6, for the breakdown, from the same source: elements related by timbre glue a track together. It is deferred there because its only job is to hold the breakdown, and before arrangement there is no breakdown.

Hence a naming confusion: in Sampler the Texture mode is a granulator that sustains a sound indefinitely from one point, while Slice cuts the recording at transients and spreads the pieces across the keys, handy for finding a fragment in a long recording. For a stab neither is needed, Shot mode is enough (the mode selector on the right of the Sampler interface).

Length

200 to 800 ms. Shorter than 150 ms and the sound merges with the percussion; longer than a second and it stops being an event and becomes a short texture.

More precisely, count from the gap between repetitions rather than in milliseconds: the sound occupies no more than a quarter of the gap it plays in. Once per 2 bars at 128 BPM is a 3.75 s gap, so up to 900 ms.

The sense of length is better obtained from reverb than from a long sample: a reverb tail is quieter and duller than the main sound, so it fills space without demanding attention. Trim the sample short.

What to do in Bitwig

1. Record your own sound. Phone, dictaphone, tape recorder, street, skateboard on asphalt, dishes, a door. Length 5 to 20 seconds.
2. Put the file in one permanent folder with a meaningful name like 2026-08-15-metal-shelf-kitchen.wav (date first gives you sorting) and add that folder to the Bitwig browser. Turn on copying external files into the project in the project settings: a library sample can always be found again, your own recording cannot.
3. Load into Sampler, trim to length, set a 20 to 30 ms fade so the cut does not click.
4. Process (see below). Not everything at once, one or two devices.
5. Play it rarely: once every 2 or 4 bars. A rare event is perceived as a detail, a frequent one as a part. Minimal wants a detail.
6. Alternative without recording: a three note chord in Polymer with a seventh or ninth, Multi-Note to widen it, slight detune, long release, filter nearly closed. The seventh and ninth are there so the chord does not read as major or minor and does not impose a mood on the track.

Processing: order and choice

Chain order: Sampler then distortion then resonator then filter then delay then a reverb send. First what changes timbre, then what adds time. A delay before distortion means all the repeats get distorted too.

Distortion (Bit-8, Distortion). Roughens and adds high frequencies, so the element cuts through the percussion better. Mix at 30 to 50 percent so the core stays clean.

Resonator. Gives pitch to a noisy source that has none: internally it is a bank of narrow resonant filters tuned to specific notes. Set a note from the key and start the dry/wet balance at about a third. Substitute if unavailable: a band pass filter with high resonance.

Filter with modulation. A stab plays every 2 bars, so thirty times a minute, and without change that becomes obvious. LFO of 8 or 16 bars, raise resonance or the movement is inaudible. Cheap alternative: assign Velocity to the filter and vary velocity slightly in the clip.

Delay. Unlike on the ghost, Feedback is lower here, 20 to 30 percent (you want two or three repeats, not a chain), and the time is 1/8 or 1/4 rather than 3/16: here the delay supports the grid rather than drifting against it. Mix around 30 percent, the original sound is needed.

Where to start. Noisy source with no pitch: resonator plus reverb. Tonal source: filter with LFO plus reverb. Add distortion and delay as a third step, often they turn out unnecessary.

Reverb via a send

A send routes a copy of the signal to a separate track with an effect without removing the original sound. The point is that several tracks sent to one reverb end up in what sounds like the same room. A separate reverb per track gives several different rooms and the track falls apart.

1. Create a track of type Effect (right click in the track list). Sends do not appear on a regular audio track.
2. Put a Reverb on it, time 0.6 to 1.2 s. A filter before the reverb is needed too, but that is hygiene and goes in during Module 7.
3. Set the reverb’s Mix on the Effect Track fully wet. The dry sound travels on its own track; here it would only double.
4. Raise the send control on the Stab track. It is visible in the mixer and in the Inspector, labelled with the Effect Track name.
5. Level: mute the whole Effect Track and unmute it. Correct when muting makes the sound dry and flat, but with it on the reverb is not audible as a separate effect.

If the source is too recognizable

A field recording often brings a specific subject with it (a bird, a car, a door), and the track starts reading as illustration. Recognizability rests on three things: the full length of the sound, its characteristic pitch contour, and repetition. Break at least one, preferably two.

• Trim to a fragment. Take 200 to 300 ms, not from the start: the start contains the attack the sound is identified by. The middle or the end give a tonal fragment of unclear origin.
• Transpose down by 5 to 12 semitones. Many recognizable sounds live in a narrow pitch range and stop being identifiable outside it. Down beats up: with Repitch on, slowing down adds graininess.
• Remove the characteristic top with a filter. Often it is the high frequencies that carry the identifying feature.
• Do not use a delay with high feedback if the source repeats in series in nature: repetition brings recognizability straight back.

Setting a band pass filter, if you use one instead of an HP plus LP pair: work out the center between the desired edges by ear, not arithmetically. Hearing is logarithmic, so the middle between 300 Hz and 4 kHz sits around 1.2 kHz rather than 2.1 kHz.

About pitch matching. With sounds whose pitch slides or is smeared, dissonance against the bass usually does not arise at all: the ear does not treat such a sound as a note and does not compare it with the bass. If you turn Pitch a couple of semitones either way and nothing gets worse or better, the question is closed, set it by character. This distinction trains itself on the project with harmony; do not chase it here.

Exercise. Constraint: the element plays no more than once every 2 bars.

Done when: the sample or preset is saved, the clip sits on the track, and in the four element loop nothing feels like it needs adding.

Module 5. Movement

Goal: no element sounds the same way twice.

Why this is the key module

Part of it is already done in Module 2: without slow modulation and randomness the loop simply does not pass the three minute test. Here the same thing is extended over the whole track and turned into a system.

The difference between a loop and a track in minimal consists almost entirely of this. In a track with melody, attention is held by harmonic change; here there is no such anchor, so its role is played by slow parameter change. The listener does not register these changes individually but senses a direction: something is developing, so it is worth listening on.

The rule: every sounding element has at least one parameter that changes over time, and at least one change period longer than 16 bars. Periods under 4 bars the ear treats as part of the timbre, 4 to 16 as a rhythmic effect, over 16 as development. You need all three, but the long ones are what people forget.

Why modulators rather than drawn automation

A drawn line is finite: it ends where you ended it, and it has to be extended every time the arrangement changes. A modulator is set once and works always, including in bars that do not exist yet. Practically that means you can move sections freely in Module 6 without redoing anything.

Automation is left for one thing: dramaturgy, that is conscious decisions at the level of “it gets denser here”. Everything else goes to modulators.

The scope of a modulator

A modulator does not see everything. It lives inside a specific device and reaches the parameters of that device and of whatever is nested inside it. It does not reach devices on other tracks, and apparently does not reach devices that sit earlier in the chain either.

Practical conclusion: put the modulator in the same device whose parameters it should turn, or earlier in the chain. If an assignment will not take and no ring appears around the knob, this is usually why, rather than a mistake in your actions.

What to do in Bitwig

1. LFO, synced to 16 or 32 bars, on the percussion filter cutoff. Small depth, right at the edge of noticeability. Raise the resonance of that filter: without resonance a moving filter just reads as “duller and brighter”, with resonance you hear where exactly it is going and the modulation starts to work. Noticeable filter modulation reads as an effect and gets old; unnoticeable modulation reads as life.
2. Curves: a hand drawn curve over 32 or 64 bars assigned to several parameters at once. Unlike an LFO the curve is asymmetric, and asymmetric development does not read as automation.
3. Steps on percussion timbre, per step: different values on each step give the feel of a performed part without a single new note.
4. Random with smoothing on pan and on ghost pitch, minimal depth. Smoothing is mandatory, otherwise it jumps.
5. An LP filter on the Drums group. Not for modulation but because it is needed twice in Module 6: for the breakdown where the whole rhythm rides up and returns, and for the run ups where the top is cut for two bars before a new element enters. Four separate automation lanes make that painful and worse sounding, because the elements move out of sync. Put it in now, cutoff wide open, a little resonance.
6. Macro-4, with a caveat. A macro has no movement of its own, it is just a knob passing its position to several parameters at once. Because of the scope limitation it does not work as a global “density” knob for the whole track: it will not reach parameters on other tracks. Its real use is narrower: linking several parameters inside one device so you can move them with one gesture. For example filter cutoff and resonance together, so opening also sharpens, and in the arrangement that is one lane instead of two. If each track has only one parameter to automate, you do not need a macro, automate directly.
7. Automation Clips (Bitwig 6): a filter movement built once can be saved as a clip and reused in different sections, stretched and moved.

Exercise. A 64 bar loop; listen all the way through and write down what changed between the first bar and the last. Fewer than three items means there is not enough movement.

Done when: the Drums group has a filter, at least 4 modulators are working, and the 64 bar loop plays without you wanting to edit.

What to put where

Tested in practice, by track:

Perc. LFO of 16 or 32 bars on the filter, resonance up. Velocity to the same filter: the level spread from Randomize becomes a timbre spread too. Random with smoothing on pan, period 2 or 4 bars, small depth.

Ghost. Random with smoothing on pan. Almost nothing else is needed, it already changes through the delay.

Hats. Velocity on the filter. Steps on timbre: use 12 or 14 steps rather than 16, so the cycle does not coincide with the bar and the accents land in new places each bar. Here the displacement trick does work, unlike the clip length in Module 2, because a straight background already exists.

Stab. LFO on the filter. Random at small depth on pitch, within tens of cents, gives slight detuning from repeat to repeat.

Kick and Bass. Nothing. They need stability; any movement reads as a mistake.

Good things to modulate: pan, reverb send level (the ear reads distance as a natural property rather than as processing), resonance, length and Decay, pitch within cents.

Bad things: level (the ear is most sensitive to it and spots a change instantly; the one acceptable exception is overall hat level within 1 to 2 dB) and anything affecting the grid.

Two modulators on one parameter work better than one: an LFO of 32 bars plus Random at small depth on the same knob give slow directed movement with organic jitter on top.

How to set depth

Depth is set not in the modulator but by the ring around the knob it is assigned to. Universal check: switch the modulator off and listen for a minute, switch it on and listen for a minute. Correct when it is noticeably deader with it off, but with it on you cannot hear how exactly the parameter is moving. There is a wide zone between those points, precision is unnecessary.

The exception is Steps: it is supposed to be audible, that is its purpose.

About POLY. Modulators have a switch making them polyphonic, that is separate per note. Slow timbre modulation does not want that: you want one shared wave per track rather than one per hit.

Materials

• Bitwig Studio 6: new automation, Automation Clips, Clip Aliases
• Polarity on YouTube and the lesson archive: the best source on modulation and the Grid, with semantic search in the archive

Module 6. Arrangement

Goal: a 6:30 track built from the loop you already have.

Where that length and that map come from

Six and a half minutes is function, not aesthetics. A track of that length can be mixed in and out, and it has 16 bars at each end where almost nothing happens: that is the room for mixing. Hence the empty intro and outro, which feel redundant on home listening and are necessary in a set.

Second: 16 bars at 128 BPM is 30 seconds and 32 bars is a minute. Counting in bars is more convenient than in seconds, because all events land on multiples of 8 and 16 anyway.

A 208 bar map at 128 BPM (one bar is 1.875 seconds)

What to do in Bitwig, step by step

Preparation

1. Before moving anything, make sure the Clip Launcher holds the clip variants you need per track: the main one and, where needed, a sparse one for the intro. Do not invent variants while moving clips, those are different modes of work.
2. Mark the end. Bitwig has no separate end marker, the last clip defines the length, so it is easier to stretch the arranger loop from bar 1 to bar 209 (the end of bar 208 is the start of 209). The loop can be disabled and the markers stay visible; on export you can pick it as the range. If you start losing track of the map, add Cue Markers at the section boundaries.

Moving clips

3. Lay things out by track, not by section: first the kick across its whole length from bar 1 to 208, then the hats in full, then percussion, and so on. It is easier to hold the map in your head this way and you can see where each track is silent. Laying out by section makes you switch tracks twenty times.
4. Kick: bars 1 to 96 and 113 to 208 (absent in the breakdown). Hats: 1 to 96 and 113 to 208. Percussion: from 17. Ghost: from 33. Bass: from 33, except the breakdown. Stab: from 65. In the breakdown, 97 to 112, what remains is whatever sustains: the stab with a long reverb and, if you have it, the texture.
5. Clip Aliases (Bitwig 6) for repeated sections: editing one alias changes all the others. In a genre with this much repetition that saves the bulk of the work; otherwise a late groove edit means manually replacing twenty copies. Where you want a unique variation, use a normal copy rather than an alias.

Texture

6. Texture is made here rather than in Module 4, and from the same source as the stab: a related timbre glues the track together. The same file into a second Sampler, playback mode looped (not Shot, or the sound plays once and stops), loop crossfade not at zero or every pass will click. Playback point where the recording is most even. Envelope: attack 1 to 3 s, release 2 to 4 s, sustain high. Close the top to about 3 kHz so the texture does not argue with the hats. Transpose into the key: a sustained sound is heard as a note, so this matters more than with the stab. LFO of 32 bars on the filter.
7. It plays only in the breakdown: one long note from bar 93 (so it has time to swell before the kick leaves) to 116 (so it does not cut off exactly on the return). Resist the urge to run it through the whole track: only something that has not been heard before can hold a breakdown. If the texture runs from the start, the breakdown is just the same thing minus the kick, and the drop feels like a hole rather than a change of state.
8. If the texture appears and disappears abruptly, check in order: the envelope (usually the stab settings are still there), the note length, the clip boundaries (the release gets cut by them), and as a last resort draw level automation.

Automation

9. Open the automation lane with A. To automate a device parameter it is easier to turn the knob than to hunt through the list: Bitwig will put it into the selector for you.
10. Lane 1, the filter on the Drums group. The main lane of the track. Bar 1 half closed, bar 33 opens almost fully, bars 81 to 96 open (first peak), bar 97 sharply down over a bar or two (a slow close feels like a fade rather than a decision), bars 97 to 112 down, bar 113 back up but not to maximum, to about three quarters, bar 129 fully open, from bar 177 gradually down, by 208 roughly where it started.
11. Lane 2, ghost level. Quiet until bar 32, working level from 33, out for 97 to 112, back at 113, out at 193. Ghost is the least noticeable element, so its arrival and departure change density without drawing attention to themselves.
12. Lane 3, stab reverb send. 65 to 96 a small amount, 97 to 112 raised a lot (in the breakdown the reverb occupies the space the drums vacated without adding new elements), 113 back, 177 to 192 raised again.
13. Draw one lane at a time and listen after each. Three at once cannot be judged. Do not automate kick and bass level, or parameters that already have modulators on them: you get a conflict.

Transitions

14. The job of a transition is not to decorate but to warn the ear that something is about to change, a bar or two before the event itself. Without a warning even a correctly placed event sounds like a splice.
15. Four methods by strength: cutting the hats for a bar before the event (the mildest, done with scissors on the clip); the Drums group filter down for 2 bars (stronger, because everything moves at once, so use it less often); a pause of half a bar (the strongest, exactly once per track); a reverse of a sample ending on the downbeat of the target bar (gives a rise toward the point).
16. Where to put them specifically: bar 32 cut the hats, bar 64 cut the hats, bars 111 to 112 pause, bar 128 reverse. Bars 96 and 176 get no transition: the breakdown arrives by itself through the kick leaving, and the outro should start unannounced.
17. A method only works relative to the current state. Pulling the filter down before the kick returns is pointless, it is already down from the breakdown lane. Instead, run the filter up through the last two bars of the breakdown, so the return meets an already opening sound.
18. About the pause: everything active is removed while anything sustaining (the texture with its reverb) is left to decay, which makes the pause less abrupt. Half a bar, so about a second. A full bar is stronger, but on a dancefloor two seconds of silence already feels like a stop.
19. The one change rule: exactly one event every 8 or 16 bars. Two simultaneous events are counted by the ear as one, so half the work is wasted and material for later sections is spent twice as fast. If two automation lanes converge at one point, separate them.

Peak and outro

20. The peak (129 to 176) differs not in content or level but in how long the state stays still. Every previous section lasted 16 or 32 bars; this one lasts 48 and nothing enters or leaves. Until now the listener got a new event every half minute; here the flow stops and a sense arises that the track has settled into a stable state. Those ninety seconds are what Module 5 was for: without modulators, 48 bars without events is simply boring, with them it is a state. Do not add anything “to strengthen it”: the peak is strengthened by doing nothing for longer than expected.
21. If, after listening all the way through from the start, the peak still sags, add one small event around bar 153: drop the hats for 4 bars or raise the stab send. One, not two. Judging the peak by starting playback at bar 129 is useless.
22. The outro (177 to 208): elements leave one at a time, in reverse order of arrival. 177 the stab leaves, 185 the ghost, 193 bass and percussion, 193 to 208 kick and hats remain. The last thing that arrived is what the listener remembers best, so its departure reads most clearly, and by the end you are left with what the track started from, which closes the form. Departures need no transitions: the element simply stops playing, and that is fine.

Exercise. Listen to the track once all the way through without stopping and without editing, noting the timings where it got boring. Only fix those places. Listening without stopping is essential: edits made on the fly always improve an individual bar and almost always damage the form.

Done when: the track plays start to finish, the length is within the limits, and a version is saved marked arrangement done.

Materials

• MusicRadar on transitions built from bringing elements in and out
• A breakdown of typical techno track structure to check your map against

Module 7. Mix

Goal: a track that reads equally well on a phone, on headphones and on a big system.

Before you start

Rest your ears. Mixing straight after arranging almost always goes badly: you have heard the track dozens of times and stopped hearing it as a whole. An hour at minimum, a day is better.

Work at moderate volume. Decisions made loud are usually wrong.

The order of steps is not arbitrary: each one depends on the previous. Setting a compressor before levels are set is pointless, because changing the level changes all of its behaviour.

All the frequency cuts and all the checks are gathered here too. Modules 1 through 6 deliberately have none: there you decide what plays and when, and cleanup scattered along the way distracts from that and makes you fix problems that have not arisen yet. One exception: filters that work as timbre rather than cleanup (the LP inside a synth, the band pass on the stab for unrecognizability) stay in their own modules, because those are decisions about sound.

Step 1. Level balance

1. Take the limiter off the master for the duration of mixing.
2. Pull all tracks down to zero.
3. Bring up the kick. The absolute level does not matter, the ratios do; a safe starting point is the kick alone peaking around minus 10 dB on the master.
4. Add bass so that kick plus bass sounds like one thing.
5. Then one at a time: hats, percussion, ghost, stab, texture. Raise each until it is clearly audible, then back off slightly.

Balancing everything at once is impossible: changing one changes the perception of all. One at a time is the only method that works.

Reference points for minimal: kick and bass loudest, hats and percussion noticeably quieter than feels right, ghost at the edge of audibility, stab about level with the percussion.

Careful with automated tracks. A level automation lane overrides the fader entirely rather than offsetting it, so anything you set by hand is lost. The ghost has such a lane. Balance it either by moving the lane’s points or with the Gain knob on a Tool device.

Step 2. Headroom on the master

1. Play the densest part, bars 129 to 176.
2. Watch the master meter. Peaks should be around minus 6 dB.
3. If louder, select all tracks and move one fader: the rest follow and the ratios are preserved.

Headroom is needed because summing tracks produces unpredictable peaks, and you want to work where nothing is hitting the ceiling.

Step 3. Frequency hygiene

All the cuts happen here, in one pass, not during composition. The reason is that these are not creative decisions but cleanup: from each element you remove what it should not have in the first place. Scattered across the modules, these cuts distract from form and make you fix problems that do not exist yet.

Zones. The kick gets roughly 40 to 65 Hz, the bass 70 to 150 Hz, percussion 200 Hz to 3 kHz, hats everything from 3 kHz. There should be nothing left to add up down low.

Low cuts (HP), by track:

• Kick: 30 Hz. Nothing useful below, but energy goes there and eats headroom.
• Bass: 45 Hz. A separate device after the distortion, because distortion adds low end of its own. Resonance strictly at zero: a resonant bump here would land right on the kick, so the filter would be cluttering the very space it was put in to clear.
• Hats: 300 to 500 Hz.
• Percussion: 150 to 300 Hz.
• Ghost: 300 Hz.
• Stab and texture: 300 Hz (on the stab this is probably already done by the band pass from Module 4, where it was there for unrecognizability).
• Ghost delay: low cut inside the feedback loop so the repeats do not accumulate low end.
• Reverb Effect Track: HP at 400 Hz before the reverb. A reverb smears everything over time, including low end, and low end smeared over a second turns into rumble and clogs the kick’s space.

High cuts (LP) are needed less often: on the texture down to about 3 kHz so it does not argue with the hats, and on the bass after distortion if it produced ringing.

If something sounds muddy and the cuts did not help: find the offending band with a narrow band boosted hard, sweep until it becomes especially unpleasant, then turn the boost into a cut.

Subtraction beats addition: addition raises total energy and moves you toward the ceiling, while only removal frees space in the mix.

Do not get carried away. If nothing is in the way after the cuts, stop there. Hunting for problems that do not exist makes a track thin.

Step 4. Mono and phone check

Tool on the master, collapse to mono, listen.

If the low end vanished, look for stereo processing on bass or kick. If percussion vanished, two elements are panned to opposite sides and cancel in mono; move one closer to the center.

Club systems often sum the low end to mono, and stereo processing on the bass turns into lost level there. Switch mono off after the check.

Phone in ten seconds: an HP filter on the master pushed up to 300 Hz. That is approximately what a phone speaker hears. The bass should remain legible through its harmonics. If it vanished, go back to the waveform and distortion from Module 3. Switch the filter off after the check.

Step 5. Depth

Make sure the reverb is not audible as a separate effect: muting the Effect Track should change flatness rather than remove an echo.

Kick and bass should have no reverb at all.

A distant sound differs from a close one less in level than in dry to wet ratio and missing highs, so depth is cheaper to make with sends than with faders.

Step 6. Bus compression (optional)

A compressor on the Drums group: Ratio 2:1 to 3:1, Attack 10 to 30 ms, Release auto or around 200 ms, Threshold lowered until you get 1 to 2 dB of reduction on peaks, Makeup raised just enough that switching the device on and off does not change the level.

Attack is the one parameter here where a mistake is immediately audible: the kick’s click lasts a few milliseconds, and a fast attack will squash it, taking away the kick’s legibility on small speakers. Release has to let go before the next hit: at 128 BPM the gap between kicks is 469 ms, so going beyond 200 ms is unwise or the compressor will be pressing constantly.

The job of this step is to glue the drums into one layer instead of three independent ones. One to two dB sounds oddly little, but that is exactly what gluing takes; more starts changing character rather than joining.

If an honest level matched comparison shows no improvement, switch it off and move on. That is not a gap in the work. A separate compressor on the kick is unnecessary: it already has a clean envelope.

Step 7. Reference comparison

20 seconds of the reference, 20 seconds of yours, at matched level and no louder than yours. Even one dB of difference completely breaks your judgement.

Note not “worse or better” but something specific: mine has less low end, mine has brighter highs, my percussion is louder. Only fix what you were able to put into words. A feeling that “theirs is somehow better” is not a task.

Step 8. Other systems

Export a draft and listen on a phone, on laptop speakers and on headphones (or on speakers if you mixed on headphones).

Phone: does the groove read without low end. If the bass vanished, go back to the harmonics from Module 3. Laptop: are the highs too harsh. Headphones: clicks, noise, cut offs you could not hear before.

When to stop

When after another edit you cannot say whether it got better or just different. That is the sign that your hearing is tired and further work is random.

Done when: the three checks are passed, the problems you named in words are fixed, and no more edits are being made.

Module 8. Master and release

Mixing is work on individual tracks; mastering is work on a finished stereo file where individual tracks no longer exist. Hence the main point: mastering does not fix a mix. If the bass is burying the kick, that is unfixable on the sum, because you cannot lower one without touching the other. If something sounds bad at this stage, go back to Module 7 rather than pushing the limiter harder.

In practice mastering does three modest things: brings the level to an accepted standard, removes small frequency imbalances that only show on the sum, and sets a ceiling so encoding does not produce distortion. On a first track the difference between good and mediocre mastering is negligible next to the difference between a good and a mediocre mix.

Step 1. Limiter

A limiter is a compressor with an infinite ratio: it simply does not let signal past the ceiling.

1. Put the Peak Limiter back on the master.
2. Ceiling at minus 1 dB. Not zero: MP3 and AAC encoding changes the waveform and peaks grow, so a track mixed right up to the edge will distort on streaming.
3. Raise the input gradually and stop at 2 to 3 dB of reduction in the densest places.
4. Check: remove the limiter and compare at matched level. If with the limiter the track went flat, the kick lost its punch and the percussion squashed, lower the input. The feeling that “louder means better” is a trap.

Step 2. Loudness

Integrated loudness of minus 10 to minus 8 LUFS for a club track. Without a LUFS meter, use the reference: your track should sound about as loud at the same knob position.

Chasing more is pointless: streaming normalizes to its own level, so an overly loud master simply gets turned down having lost its dynamics. On a club system the DJ sets the level, and there the more dynamic track wins.

Step 3. Export

1. WAV, 24 bit, 44.1 kHz. No dithering needed at 24 bit.
2. Range from bar 1 to 209 or by the loop. Check that the tail of the last sound fits, extend by a couple of bars if needed.
3. A second file as MP3 320 for sending.
4. Open the exported file in a player, not in Bitwig, and listen to the start and the end. Cut offs, clicks at the boundaries and missing tails only show up this way. Also check there is no extra silence at the head.
5. Name the file with the title and the date: in six months that is the only thing that will let you find the right version.

Step 4. Publishing

This is part of the module, not an option.

1. Upload to SoundCloud or Bandcamp.
2. Write a paragraph on the microblog: what it is, how it was made, what turned out to be unexpected.
3. Post the link.

An unfinished last mile turns a finished track back into an unfinished project: a file on a disk nobody knows about feels psychologically like a draft, and the next project starts from the same state as the previous one.

About not liking the track. You probably will not, and you publish anyway. Your assessment of your own track at the moment of finishing has almost nothing to do with its quality: it depends on how many times you have heard it, and you have heard it more than anyone. In a month it will sound different. And the criterion here is not “a good track” but “a closed track”: the first depends on experience, the second on a decision.

Done when: the link exists.

Appendix 1. What to listen to

Classics and roots:

• Robert Hood, Minimal Nation
• Plastikman, Consumed
• Basic Channel and Maurizio: how dub minimal works
• Ricardo Villalobos, Alcachofa
• Akufen, My Way: microsampling

Contemporary and Romanian minimal:

• Raresh, Rhadoo, Petre Inspirescu: the [a:rpia:r] compilations give a sense of the genre faster than individual tracks
• Priku, Barac, Cristi Cons
• Sedef Adasi, Vril, Efdemin: the more technical and dubby edge

Listen not for inspiration but for measurements: how many elements, when they enter, what changes.

Appendix 2. Standing resources

• Bitwig Learn: official videos and articles, including the Let’s Build a… series on the Grid
• Polarity: the largest archive of Bitwig lessons, courses, device guides, project files
• Polarity: a generative techno patch in the Grid, if after the first track you want to go generative
• Bitwig Studio 6, what is new
• Bitwig’s built in documentation: the Help section with interactive lessons, worth going through once in full

Appendix 3. Common mistakes

1. Adding an element instead of improving an existing one. Boredom in minimal is almost always cured by subtraction or by changing one parameter over time.
2. Starting to mix before arranging. A loop balance does not transfer to a track, because in a track the elements sound in different combinations.
3. Listening to references louder than your own track. Louder always sounds better, which is a property of hearing rather than of quality.
4. Making kick and bass separately. They share an octave and can only be set together.
5. Endlessly browsing samples. Hence the ban on new libraries after Module 0.
6. Drawing automation where a modulator would do. A drawn line breaks with every arrangement edit.

Appendix 4. If the genre turns out to be too subtle

Minimal is paradoxical: it is technically easy to make (a sine, a noise hat, three bass notes) and hard to judge, because the criterion is subjective and needs a discriminating ear. The risk is not that it will fail, but that it will come out exactly per instructions and be boring, with nothing to tell you what to fix.

What to do about it without changing the course: shift the track toward dub techno or ambient techno. Same skeleton, same modules, plus a chord stab with a long reverb. Reverb forgives imprecision, and a chord supplies emotion you would otherwise have to squeeze out of micro timing. Module 4 already allows this, there is a route through Polymer with a seventh.

The second remedy is the ear training section at the start. Those three distinctions remove most of the problem.

Appendix 5. Protocol for getting stuck

1. A 25 minute timer, working only on the point of the module where you stopped. Do not skip forward and do not go back.

2. If after 25 minutes you do not like the result, save and commit anyway. A bad committed decision beats a good uncommitted one, because it lets you move, and you can return to it from the next track.

3. If a week got skipped, do not start over. Open the last version, hit play, and change nothing for the first 5 minutes.

4. If you do not like the material at Module 3 or 4, move on anyway rather than improving what is there. Three reasons: at this stage about half the track is sounding and the foundation is supposed to sound bare; some of the complaints will disappear by themselves after Modules 4 and 5 with no edits at all; and specific complaints only become formulable after Module 6, when you can hear the track as a whole. Before moving on, write three lines about what exactly you do not like, and close the file. Written down, it waits until Module 7; unwritten, it pulls attention through the whole next session.

   The feeling of “I do not like this” in the middle of a first track happens to almost everyone and almost always at the same spot: the technical part already takes effort and the result has not started rewarding you yet.

5. If the track seems bad at Module 6 or 7, take it to release anyway. A judgement made inside the process means nothing: it depends far more on how many times you have heard it than on the quality of the material.