Patch Portal: how the sounds of late-night electronic music get made
A reference and playground for synthesis. 64 sounds, each a step-by-step recipe you can build in any synth. This text version is for readers and search tools without JavaScript; the full interactive site loads with JavaScript on.
How to make a cold, held ambient pad, the bus-stop chord
Bus Stop Pad. Held chord at the temperature of cold concrete.
The bus stop light is always on, even when nobody's coming. This pad sits in that glow. Five soft-saw wavetable voices, detuned just enough to flicker, held under a low filter.
What you'll need: You'll need a synth with a unison (voice-stacking) control, a lowpass filter with an LFO routed to its cutoff, a noise layer for air, and reverb. All standard.
The recipe
- The wavetable is the light itself. Start with OSC A, the first oscillator, on a soft saw wavetable, one that doesn't bite in the top harmonics. You want something that could pass for a streetlamp humming when nobody's listening. You'll know the wavetable is right when the raw tone sounds dull and a little tired on its own, with no glassy fizz riding the top.
- Five fluorescent tubes, slightly out of sync. Set unison to five voices, detune twelve cents. Not a thick unified stack, more like five fluorescent tubes in five slightly different states of aging. Aim for a slow shimmer crawling through the held note rather than a single fat tone; a fast beating or chorused quality means the detune has gone too wide, since wider detune widens the gap between voices and the gap sets the beat rate.
- Keep the top out of the parking lot. the 24 dB/oct ladder lowpass on the filter, the steep kind that takes the top off fast, cutoff around 680 Hz. Low resonance, you're not drawing attention to the filter, you're just deciding what belongs outside. Done right, the pad becomes a warm glow with no edge in the top and the filter stays inaudible; a whistle or peak at the cutoff means the resonance has crept up.
- The flicker you only notice if you stop and look up. LFO 1 on the cutoff, very slow, around 0.08 Hz. Small depth. Not a pulse. A breath. You'll know it is right when you only notice the movement if you stop and wait for it; the moment you can tap along to the flicker, the depth or rate has gone too far.
- Already there before the key. Long attack, long release. 1.2 seconds in, 2.4 seconds out. The pad should feel like it was already there when you arrived. You'll know the envelope is right when you cannot catch the exact moment the note begins, and it keeps glowing for a breath after you lift your hand.
- The detail that makes it outside. A whisper of pink noise, filtered to about 3200 Hz, level at four percent. This is the thing that translates the sound from 'inside somewhere' to 'the specific outside that's cold.' Mute it and the pad feels sealed indoors; unmute and a faint cold air arrives that you sense more than hear. If you can pick the hiss out as its own layer, it is too loud.
Play it in the lower-middle of the keyboard. Hold a chord. Listen to the flicker. In a mix it lives low and behind, under the drums and the voice, filling the floor without reaching for the front. A Highway Pad underneath, two notes an octave apart, lends its idling sub to warm the floor this one only fills.
Variations:
- Push the LFO depth until the flicker becomes a pulse. Now it's a different bus stop, the one in a worse part of town.
- Add a short delay with 30% feedback. Now the bus stop is under a bridge.
In the lineage of:
How to make a warm idling-car pad that fades into static
Highway Pad. The glow of an idling car when the radio has faded to static.
You're not going anywhere. The engine is on because stopping would mean you've decided. Three detuned triangle voices over a sub octave, a wide warm pad that idles in place, going nowhere, on purpose.
What you'll need: You'll need a synth with a unison control, a sub or second oscillator for the low octave, a lowpass filter, and an amp envelope with a slow attack and release. All standard.
The recipe
- A triangle-based wavetable, slightly warped. OSC A on a soft triangle, with a gentle waveshape bend at 30%. You want the shape of something analog that's been sitting in warm air for a while. You'll know the bend is right when the tone gains a soft hollow body without ever sounding buzzy or reedy.
- Three voices, not five. Unison at three, detune around eight cents. This is a small pad, not a wide one. The dashboard is a small space. You should hear a gentle warmth and a hint of slow beating, not a spread; if the stereo opens up wide, you have used too many voices or too much detune.
- The tape is running. A slow LFO (0.12 Hz) on OSC A's fine tuning, plus or minus six cents. This is the wow-and-flutter that says tape, not digital: real tape wavers as transport speed drifts, where digital holds pitch locked. A held note should sag and recover in pitch like worn tape; once the wobble becomes a clear vibrato you can sing, the depth is too high.
- Low, close, warm. the four-pole ladder lowpass, a steep one, at around 900 Hz, resonance at zero. The cutoff is set to remove exactly enough air to feel enclosed. You'll know the cutoff is right when the pad sounds like it is heard from inside a closed car, close and muffled, with the open-window brightness gone but the warmth intact.
- The engine underneath. Enable the Sub oscillator on sine, minus one octave, level at 20%. This is the hum of the idling engine, barely audible, entirely structural. Set right, you feel the floor firm up more than you hear a new note; if you can name the sub's pitch over the pad, it is too loud.
- Slow in, patient out. Attack 800ms, release 1800ms. Not a pad that sweeps in, one that was running before you noticed and continues after you leave. There should be no audible onset to point to, and the note keeps idling for nearly two seconds after you release.
Play two notes an octave apart. Let them sit. Don't add a third. It wants the center of the mix, close and narrow, so leave the width to everything around it. Let a Bus Stop Pad supply that width, wide and low where this one stays narrow, the two pads sharing one chord at different distances.
Variations:
- Add a touch of chorus at the end of the chain, the dashboard is now behind a rain-streaked windshield.
- Bring the sub level to 40% and the pad becomes the engine more than the glow.
In the lineage of:
How to make a fluorescent-hum drone with electrical buzz
Fluorescent Drone. Sustained tone with electrical hum baked in.
A struck glass bowl that never stops ringing. FM at a non-integer ratio gives it metallic, faintly out-of-tune partials (a fluorescent tube's hum, held and magnified), and an envelope that never lets go. It doesn't resolve; it just sustains. For ambient, soundtrack, the quiet hours.
What you'll need: You'll need a synth that can do FM, one operator modulating another at a non-integer ratio (that ratio is what gives the metallic shine), a slow LFO on the amplitude for the swell, and an amp envelope that can hold a note indefinitely. Most synths have all of these.
The recipe
- A triangle, low. OSC A on triangle, played around C2. The triangle gives you odd harmonics without the aggression of a saw. You'll know it is right when the low tone reads as round and a little hollow, with none of the buzz a saw would put on top.
- A second voice, an octave up, at half volume. OSC B on triangle, one octave above, level at 50%. This is the buzz of the panel lights above the compressor. You should hear a thin upper sheen sit on top of the low tone; if the octave starts to sound like a separate held note, pull its level back.
- The compressor cycle. LFO 1 at 0.06 Hz on amp level, depth 25%. This is the slow on-off of the refrigeration, the rhythm of the drone. The volume should rise and fall over roughly fifteen seconds, felt as a slow tide rather than a tremolo you could count.
- Narrow, focused, present. Filter 1 as a lowpass at 1200 Hz. Keep it still. Keep it close. Done right, the drone sits narrow and present in the mid-range, with no sparkle up high and no sense of distance.
- A thin layer of FM. Add an FM modulator to OSC A at a non-integer ratio, near 3.5 to 1, depth 12%. The off-the-grid ratio is what tilts the partials out of tune, the faint metallic edge of the compressor housing, not the compressor itself. You'll know the depth is right when the tone gains a faint inharmonic shimmer at the edges; if it starts to ring like a bell or sound clearly detuned, the index is too high.
- Never attacks, never releases. Attack 50ms, release 50ms, but the LFO on amp is doing the envelope work. This is a sound that is simply on. The tell is notes that neither swell in nor fade out on their own; the only movement you hear is the slow compressor cycle of the LFO.
One note, held. Let it ring and the reverb bloom. The sound is in the sustain, not in any change. It works as a bed under everything, constant and low, with a little scooped out around 200 to 400 Hz so the kick and bass still read. A Streetlight Pad floated far back above it lets the drone hold the floor while the pad colors the air over it.
Variations:
- Nudge the FM ratio further from a whole number and the partials grow stranger, more bell, less tone.
- Shorten the reverb and the glass moves indoors; stretch it past four seconds and the drone becomes the room itself.
In the lineage of:
- Alessandro Cortini (Sonno)
- Stars of the Lid (The Tired Sounds of Stars of the Lid)
How to make a warm sodium-streetlight pad
Streetlight Pad. That orange light doesn't flatter anybody. It's honest.
The color of light that tells you what time of night it is without a clock. A still, far-off pad, filtered toward a color rather than a note.
What you'll need: You'll need a synth with a unison control, a sub or second oscillator, a lowpass filter with an LFO on its cutoff, plus a noise layer and reverb. Most synths have all of these.
The recipe
- A sustained low tone, the orange itself. OSC A on saw, played at C3. Level at 100%. The saw carries enough harmonic body to translate the density of the light. You'll know the source is right when the raw saw sounds bright and dense before any filtering, with plenty of upper harmonics for the lowpass to carve away.
- Heavy filter, warm curve. the 24 dB/oct ladder lowpass, the steep kind that takes the top off fast, at 480 Hz, resonance at 10%. You're filtering toward a color, not filtering out frequencies. The ladder curve is what makes this orange rather than yellow.
- A high-end haze. Noise oscillator, pink, bandpassed around 8kHz, level at 3%. This is the visible particulate in the light cone, the air made visible. You'll know the level is right when a faint high haze hangs above the dark pad and disappears the moment you mute it; if you can hear it as hiss in its own right, it is too loud.
- Slow harmonic drift. LFO 1 on filter cutoff, rate 0.04 Hz, depth 15%. Very slow. The sodium lamp's color temperature isn't static; it shifts over the course of the night. The warmth should seem to deepen and lift across the better part of a minute, never as a sweep you can hear starting or stopping.
- A sub that grounds it. Sub oscillator on sine, minus two octaves, level 30%. Keeps the sound rooted, the streetlight is attached to a pole attached to the ground. The tell is the pad feeling anchored and weighted at the bottom rather than floating; if the sub booms or muddies the low end, ease it back.
- Long in, long out, never quite arriving. Attack 2 seconds, release 4 seconds. This is not a sound that starts. The pad fades up so gradually that you miss its arrival, and it keeps decaying for several seconds after you let go.
A single low chord, held. Close your eyes. Think about the color. In a mix it sits far back and low, more color than note, so it never competes with what is up front. Set a Bus Stop Pad close in front of it, near and dry against this one's distance, so the same chord reads as both the far light and the near one.
Variations:
- Drop the cutoff to 400 Hz and the lamp is fogged over.
- Raise the noise layer to 8% and the air itself becomes the instrument.
In the lineage of:
- Bibio (Vignetting the Compost)
- Actress (Hazyville)
How to make a warm, slightly detuned bedroom Rhodes
Cheap Rhodes. Warm and drifting, a few cents shy of in tune, and better for it.
Heard through the wall from the room next door, an electric piano gone soft and a few cents out of true. The Rhodes is the easy part; the lesson is what the room and the cheap gear do to it.
What you'll need: You'll need a synth that can do FM, one oscillator modulating another, with an envelope on the FM amount, plus a lowpass filter, a little saturation, and chorus. Most FM-capable synths cover all of these.
The recipe
- A two-operator FM pair, the canonical electric-piano ratio. OSC A sine carrier, OSC B sine modulator at 14:1 ratio, the ratio behind almost every FM electric-piano patch of the era, a high ratio that throws the FM sidebands far above the fundamental into the metallic tine register the ear reads as a struck Rhodes. Modulation index around 0.5–0.7, enough harmonic content to read as electric piano, not so much that the bell-pop dominates.
- A tine envelope on the modulation depth. ENV 2 → FM modulation depth. ATK 1ms, DEC 120ms, SUS 0%. The 'tine pop' at note onset, the thing that separates a Rhodes from a sine pad. Each note should open with a brief bright bark that settles into a soft sustain within a tenth of a second; if the bright part lingers, shorten the decay.
- Slow pitch LFO, a tuning that will not sit still. LFO 1 on OSC A fine tuning, sine at 0.14 Hz, depth ±8 cents. Slow, irregular, never quite resolving, the drift of an instrument that was never quite in tune. You'll know it is right when a held note drifts gently sharp and flat as if the tuning is breathing; if the wobble speeds into a vibrato you can feel as pitch, the rate or depth is too high.
- Gentle saturation. Soft saturation at 30%. Warmth and a slight compression of the transients, the warmth coming from the extra harmonics saturation grows above each note. Listen for the tine pop sitting a touch softer and the body warmer and fuller; if the tone turns gritty or buzzy, you have driven it too hard.
- Lowpass at 3 kHz. 24 dB/oct lowpass at around 3 kHz. Rolls off the digital edge so the Rhodes feels recorded, not played live.
- Sit it in the mix. High-pass at 80 Hz to remove sub rumble; 2:1 bus compression with a 30 ms attack so the tine pop survives; gentle saturation across the master to glue it together. The chain is what makes the Rhodes feel recorded, not played live.
Play a slow melody, the patch carries the drift even when you don't play anything. In a mix it lives in the middle, so ease a little around 300 Hz if it crowds a vocal, and let the detune keep it from sounding too clean. A Warped Lead traced above it, a single drifting line answering the chords, carries the same out-of-tune ache up into the melody.
Variations:
- Bypass the pitch LFO and the patch becomes a clean, stable Rhodes, recognizable, but the drift is gone, and with it most of the hazy, dreamlike quality this patch leans on.
- Raise the modulator ratio to 14.7 and the bell-pop becomes audible at note attacks, closer to a stock FM patch, less to a recorded Rhodes.
In the lineage of:
How to make a polyrhythmic FM pluck from three interlocking clocks
FM Pluck. Three plucks, three clocks, one melody.
When no part is the melody, the pattern they form together is. Three short FM plucks on different clocks, interlocking into one line.
What you'll need: You'll need a synth that can do FM with an envelope on the FM amount, a short plucky amp envelope, a sequencer or arpeggiator for the cross-rhythm, and reverb. Most cover all of these.
The recipe
- Three FM plucks at different ratios. Three voices: ratio 1:3, ratio 1:5, ratio 1:7. Each is a short FM pluck with modulation index ~0.7. The different ratios give each voice a distinct timbre, because a higher modulator ratio spaces the FM sidebands wider and higher above the carrier. You'll know it is right when, played one at a time, you can tell the three plucks apart by ear, from rounder to brighter and more hollow as the ratio climbs.
- A pluck envelope, same on all three. AMP ATK 2ms, DEC 250ms, SUS 0%. Brief plucks, long enough to ring, short enough that overlap is sparse. You'll know it is right when each note has a clear attack and a short tail that has mostly died before the next one lands; if the plucks smear into a wash, shorten the decay.
- Different cycle lengths. Voice A repeats its arpeggio every 4 beats, Voice B every 5, Voice C every 7, each free-running across the barline rather than resetting on the downbeat. Sharing no common factor, 4, 5, and 7 only realign every 140 beats (4 times 5 times 7), so the line they form is fractal, not melodic. The combined pattern should never seem to repeat over a long listen, and no single voice ever steps forward as the tune.
- A long reverb behind them. Reverb decay 3 s, wet 45%. Holds the plucks in shared space without blurring them.
- Glue the three voices. Each voice on its own bus with its own short reverb (different decay times, 2 s, 3 s, 4 s), panned ±30% / center / ∓30%. A bus compressor across all three glues them into one music while keeping the timbres readable; the polyrhythm is in the pattern, the cohesion is in the master bus.
Let the three voices run. After a minute you can’t tell which is the line. In a mix these plucks want a little space and delay, sitting above the pad and under the vocal so the interlock reads. Thread a Music Box Pluck through the lattice on a fourth, slower cycle, a softer voice that surfaces now and then where the three cross.
Variations:
- Synchronize all three voices to the same cycle length and the patch becomes a regular arpeggio, the lattice lesson collapses to a melody.
- Add a slow filter LFO across all three voices and the lattice gains evolution, closer to Plaid than to early Autechre.
In the lineage of:
How to make a single-oscillator mono synth lead
Mono Synth Lead. A single oscillator carrying the song.
One note, sung half-awake into a room that is mostly reverb. A single oscillator, made hypnagogic by the chain it sits inside.
What you'll need: You'll need a synth with a mono (single-voice) mode, a lowpass filter, an amp envelope, and reverb. Nothing unusual.
The recipe
- A single sawtooth, mono. OSC A on saw at E4. Single voice, no detune, no unison. The source is intentionally plain, the patch lives in the FX. Solo it dry and you'll know it's right when it sounds almost boring, one thin reedy saw with no width or sparkle of its own.
- A slow-pluck envelope. AMP ATK 30ms, no decay, full sustain, REL 500ms. Slower than a pluck, faster than a pad, sitting between intent and atmosphere. Each note should ease in without a click and linger a breath after you let go.
- Lowpass at 3 kHz. Gentle lowpass with a touch of resonance (0.1). Just enough to soften the saw’s top edge, since a sawtooth carries every harmonic and the buzz is the highest of them, which the lowpass thins away. You should hear the buzz come off the top while the note still reads clearly; go too far and it turns muffled, dull rather than soft.
- Chorus, slapback delay, long reverb. Chorus depth 40%, slapback delay around 180 ms (single tap, no feedback), long reverb at 4.5 s decay, 55% wet. The chain is the patch, bypass any element and the John Maus signature disappears.
- Slip it under a voice. Sidechain the patch lightly against the vocal so it ducks 1–2 dB when she sings; a small high-shelf cut at 7 kHz to keep the saw warm; mono, the FX chain is wide enough that adding stereo to the source dissolves the patch.
Play single notes, the lesson is restraint. The chain carries everything. In a mix it sits up front but soft, one voice with room around it, so keep the other midrange thin while it sings. An Underwater Lead can answer it from below, the dark filtered foil to this dry, up-front voice.
Variations:
- Bypass the slapback delay and the patch becomes a wet pad; the hypnagogic-pop signature was in the slap.
- Add a 10-cent detune by enabling unison 2 voices and the patch becomes slightly impressionistic, closer to Ariel Pink than to John Maus.
In the lineage of:
- John Maus (Bennington)
- Ariel Pink (Round and Round)
- Molly Nilsson (Hey Moon)
How to make a concrete underpass sub-bass rumble
Underpass Sub. The hum the concrete gives back when a car goes over.
The hum the concrete gives back when a car passes overhead, low and almost still, wandering a few cents as it holds. A two-voice sine sub an octave down, lightly detuned, with a slow pitch drift standing in for tape wow.
What you'll need: You'll need a synth with a sub or second oscillator, a lowpass filter, an amp envelope, and a touch of saturation. Most cover all of these.
The recipe
- A sine sub, two voices. A sine voice around A1 with a sub an octave below it. Two voices, no more. The patch is almost entirely fundamental, with nothing bright to give it away. You'll know it's right when it's felt more than heard, a round low weight with no edge or grain riding on top.
- A hair of detune. Detune the two voices about six cents apart, so the two pitches beat against each other at their difference frequency, a slow swell in amplitude. Just enough beating to feel alive, not enough to read as two notes.
- The wow. A very slow LFO on tuning, around 0.07 Hz, only a few cents deep. The pitch wanders the way a worn tape player wanders. This drift is the whole patch. You'll know it's right when you can't quite catch the pitch moving, only sense that the hum is never perfectly still; go too far and it warbles like a seasick note.
- Lowpass near the floor. Lowpass around 320 Hz, almost no resonance. Everything above the hum belongs to the street, not to the bass. Nothing should click or hiss at the front of the note, just a warm low body with no air left on top.
- Arrive slow, leave slow. Attack around 240ms, release around 1.6 s. It comes in like the car arriving and goes out like the car leaving. Right means no thump at the start and no hard stop at the end, the hum swells up and drains away.
- A little weight. Gentle saturation, around 18 percent, so the hum reads as weight on a small speaker without losing the sub on a large one. The saturation grows harmonics above the fundamental, and a small speaker reproduces those even when it cannot move the sub itself.
Hold one low note for a long time. Let the wow be the only thing that moves. In a mix it owns the bottom alone, kept mono below 100 Hz so it stays solid, with the other parts cleared out of its way. A Highway Pad over the top gives the hum a canopy, the pad carrying the harmony while this holds the floor.
Variations:
- Remove the drift and it becomes a clean, accurate sub, and a lifeless one; the wow was the patch.
- Raise the cutoff a little and some of the street comes back in over the hum.
In the lineage of:
How to make a soft, slow-changing afternoon pad
Sunday Afternoon Pad. The light changing in a way you only notice after it has changed.
This is not a sound that changes. It is a sound that is changing. A slow wavetable pad that evolves while you hold one chord.
What you'll need: You'll need a synth with a wavetable oscillator you can sweep through its frames, a unison control, a lowpass filter, and an amp envelope with a long release. Common to nearly any synth.
The recipe
- Two wavetables, chosen to morph between. OSC A with the wavetable position set to create a specific harmonic shape; this card uses a wavetable with spectral evolution across frames. Sweep the position by hand: the harmonics should open and close smoothly, no clicks or stair-steps as one frame gives way to the next.
- A very slow LFO on WT POS. LFO 1 as a smooth sine at 0.02 Hz, one cycle every fifty seconds. Depth 60%. This is the evolution; everything else is the body. You'll know it's right when you can't hear it moving in the moment, yet the chord plainly sounds different a minute later; speed it up and the motion becomes a sweep you can track, which is too far.
- A triad held across three octaves. OSC A at root, OSC B at root plus 5th, OSC C at root plus octave. The chord is built into the patch: hold one note and a chord plays. The fifth and octave sit at the simple ratios three-to-two and two-to-one, so the ear fuses them with the root. The tell is that a single key already sounds full and open, the three pitches reading as one wide voice rather than separate notes.
- Filter stays out of the way. the four-pole ladder lowpass, steep and warm, at 2400 Hz, no modulation. The motion is all in the wavetable. The filter should be doing nothing you can hear, just taking the brittle edge off the top while the wavetable keeps all the movement.
- Soft saturation, for body. Gentle tape-style saturation at 15%. The saturation gives the chord weight without muddying its evolution.
- Long envelope, no retrigger. Attack 1500ms, release 3 seconds. Enable Legato (with MONO) so that consecutive notes don't restart the LFO, the evolution continues across notes. Move to a new note and the timbre shouldn't reset; the morph carries on uninterrupted as if you never lifted your hand.
Hold one note for a full minute. The chord changes underneath without asking permission. It sits behind the track like weather, slow enough never to fight the front, so give it room and let it move. A Vocal Tape Drone slipped beneath it gives a deeper, breathing bed for the chord to evolve over.
Variations:
- Increase the LFO rate to 0.1 Hz and the evolution becomes visible, it's no longer something you notice after the fact.
- Turn off Legato and each new note restarts the evolution; the sound becomes more punctuated.
In the lineage of:
How to make a music-box pluck with a tight bell envelope
Music Box Pluck. Tight envelope, single bell tone, one note at a time.
A music box left open on a shelf, one note at a time. A bell-pluck doesn't need much; the envelope is the patch.
What you'll need: You'll need a synth that can do FM with an envelope on the FM amount, a lowpass filter, and reverb. Most FM-capable synths cover all three.
The recipe
- A two-operator FM pair, bell ratio. OSC A sine carrier, OSC B sine modulator at 3.5:1 ratio. Modulation index around 1.0–1.5, high enough that the inharmonic partials read as bell-like. The non-integer ratio is what scatters the FM sidebands off the harmonic series; the index only sets how strongly they ring.
- A pluck envelope. AMP envelope ATK 3ms, DEC 350ms, SUS 0%, REL 200ms. The note is brief, over before it has time to settle. You'll know it's right when the note strikes and falls away on its own, holding the key changes nothing; stretch the decay and it stops being a pluck.
- Highpass at 500 Hz. Highpass filter at 500 Hz to remove the body, keep only the bell-like upper register. The patch is meant to ring, not to hold weight. The low thunk drops away and only the bright ring remains, nothing left to anchor it to the floor.
- A long reverb tail. Reverb decay 4 s, wet 55%. The reverb is half the patch, without it, the pluck is a click. With it, the pluck is a room.
- Tighten the pluck. High-pass at 500 Hz to remove body resonances (the music box has no body, the reverb is its body); a 1.5 dB dip at 3 kHz if the bell reads harsh; let the reverb tail be the only thing holding the note open.
Play one note. Wait. Play the next when the room is silent again. A music box wants air around it, so keep it sparse and high in the mix, one bright point with space on every side. A Crystal Piano answering from a little lower makes a second bright voice, the two trading single notes across the silence.
Variations:
- Raise the modulator ratio to 7 and the bell becomes more glassy, less wooden, closer to a glockenspiel.
- Lengthen the decay to 1.5 s and the pluck becomes a sustained chime, the lesson shifts from envelope to harmonic content.
In the lineage of:
- múm (Yesterday Was Dramatic – Today Is OK)
- Boards of Canada (Music Has the Right to Children)
How to make a three-oscillator organ chord
Organ Chord. Press one key, three oscillators answer.
One key that answers with a whole chord, the way an old home organ does. Three oscillators tuned to root, third, and fifth; press one key, hear the triad. That's the Magnus chord-organ lesson, brought into the synth.
What you'll need: You'll need a synth with enough polyphony to hold a chord, a few stacked or detuned oscillators, chorus, and reverb. All standard.
The recipe
- Three oscillators, tuned to the triad. OSC A at root, OSC B at +4 semitones (major third), OSC C at +7 semitones (fifth). Equal volume, no voice is the lead. One key, three pitches. You'll know it's right when one key already sounds like a full major chord and no single pitch stands out in front of the other two.
- A slow-attack envelope. AMP ATK 150ms, the chord appears rather than arrives. Long release lets each held note bleed into the next as you move through changes. There should be no click at the keypress; the chord fades up smoothly, the previous chord still ringing under the new one.
- Soft chorus, hall reverb. Chorus depth 35%, reverb decay 4 s. The chord lives in a room, and the chorus widens it with short modulated delays that drift out of sync with the original, so the ear no longer places the sound at one point. The chord gains width and a tail of space around it; push the chorus and it tips into a seasick warble, which is too far.
- Hold it. Move slowly. Play the patch as held chords, one key at a time. The slow-attack envelope plus the built-in triad makes every press feel like a wash blooming, not a note sounding. Move through a four-chord progression at half-note pace and the patch carries the song.
- Layer chorus and shimmer. Dry signal to chorus, then plate reverb at 4 s decay. On a parallel bus, pitch-shift the reverb tail up an octave at 25% mix. The chord holds; the shimmer floats above it. Slight bus compression at 2:1 with slow attack so chord swells breathe between presses.
Press one key. Hear three. The chord-organ trick is hidden in the patch. It fills the midrange on its own, so let it carry the harmony and keep other pads out of its way. Underneath it, a Sunday Afternoon Pad gives a slow bed that moves while the triad holds steady on top.
Variations:
- Sequence root → third → fifth → root as an arpeggio over the bar, the same patch becomes a walking triad.
- Add an octave-up shimmer return at 30% mix and the patch leans toward Beach House’s wetter records.
In the lineage of:
How to make an underwater lead with a closed lowpass
Underwater Lead. Lowpass closed, attack softened, lead muted by depth.
The lead line was clear once. Something has happened to it since. A melodic lead pulled under a closed filter, muffled by depth.
What you'll need: You'll need a synth with a unison control, a lowpass filter with an LFO routed to its cutoff for the wobble, chorus, and reverb. Most synths have all of these.
The recipe
- A triangle lead with gentle saturation. OSC A on triangle, gentle saturation at 30%. The triangle carries the melodic clarity; the saturation gives it body underwater. The soft triangle should thicken and warm without turning buzzy, still a clean melodic tone with a little more flesh on it.
- A formant filter, the lead's mouth is closed. Filter 1 as a formant (vowel) filter, cutoff modulating between 'oh' and 'ooh' positions. LFO 1 at 0.2 Hz, moderate depth. The vocal quality comes from resonant peaks parked where the vocal tract makes those vowels, the same cues the ear uses to name one. You'll know it's right when a slow vowel opens and closes over the tone, like a voice trying to speak below the surface.
- Chorus that feels like ripple. Chorus module, rate 0.8 Hz, depth 35%, feedback 15%. Not a tight chorus, a liquid one. The tell is the edges starting to ripple and smear rather than just widen, the pitch swimming slightly under the depth.
- A wet reverb with short predelay. Plate reverb, 2-second decay, predelay 20ms, damping 40%. The lead is in the water, not near it. The dry edge of the note all but disappears into the wet, the reverb sitting right on top of it instead of trailing behind.
- Slight pitch drift. LFO 2 on OSC A fine tuning, random shape, rate 0.3 Hz, depth ±3 cents. This is the physics of the underwater environment affecting the pitch. The pitch should feel faintly unsteady, never landing perfectly in tune; deepen it past a few cents and it reads as broken rather than submerged.
- Softer attack, longer release. Attack 80ms (it's still a lead, you want to hear the note start), release 900ms. The notes take time to surface.
Play a slow melody in the middle register. No fast runs. The water doesn't allow them. Because it is already dark and filtered, it tucks under the mix easily, so place it below the vocal and let the murk read as depth. Sit a Sunday Afternoon Pad under it as a chord bed, so the lead has harmony to surface from.
Variations:
- Raise the formant modulation depth and the lead becomes more vocal, almost a word.
- Cut the chorus and the lead surfaces; the water becomes air.
In the lineage of:
How to make a soft-attack cushioned bass with no transient
Cushioned Bass. Soft attack, fat body, no transient.
Press into memory foam and watch it fill back in, slow and without edges. It's not that the bass is slow. It's that it refuses to begin sharply. A sine sub-bass with a soft octave above it, where the slow attack is the whole patch.
What you'll need: You'll need a synth with a sub or second oscillator, a lowpass filter, an amp envelope, and a little saturation. Most cover all of these.
The recipe
- Pure sine sub. Sub oscillator on sine, played around A1 or A2. Level at 100%. This is the bass; everything else is context. A sine is a single partial with no overtones, which is why it reads as pure weight. You'll know the sine is right when there is weight but nothing to grab onto, no buzz or grain, just a clean low body.
- A second layer, softer, an octave up. OSC A on sine one octave above the sub, at 40% level. This is what lets the bass be heard on a small speaker while remaining sub on a good one. A phone speaker cannot move air at the fundamental, but it can reproduce the octave above it, and the ear infers the missing root from that harmonic.
- Soft saturation, no aggression. Gentle tube-style saturation at 20%. The tube adds warmth without adding attack. Saturation grows low harmonics above the fundamental, and it is those added overtones the ear hears as warmth.
- No filter motion. Filter 1 as lowpass at 400 Hz. Keep it still. The character is in the envelope, not the filter. Nothing in the tone should sweep or open; the only thing moving is the slow swell of the envelope.
- The defining move, no transient. Attack 180ms. This is unusual for a bass. It's what gives the sound its name. The percussive strike we hear is the steep onset transient, and ramping the level over 180ms removes that edge so there is nothing to perceive as a hit. The note presses in rather than arriving.
- A long release. Release 800ms. The note recedes the way it arrived, slowly, without a line between "played" and "gone."
Quarter-note pattern, middle register. Let each note live through the gap to the next. A soft bass needs the low end to itself, so keep it mono underneath and let its slow attack sit just behind the kick. When you want grit instead of cushion in that same low seat, build Basement Bass and swap one in for the other.
Variations:
- Drop the attack to 40ms and the sound becomes conventional bass, compare to what you had.
- Add a 400ms release and the notes begin to overlap; now it's not a bassline, it's a drone.
In the lineage of:
- Burial (Untrue)
- Huerco S. (For Those of You Who Have Never (And Also Those Who Have))
How to make a vocal tape drone of adjacent, unresolving pitches
Vocal Tape Drone. Adjacent pitches that never quite resolve.
The two notes aren't harmony. They aren't dissonance either. They are next to each other. A two-voice drone holding a minor second, soft as tape.
What you'll need: You'll need a synth with a unison control, a lowpass filter, a noise layer for the tape hiss, and an LFO for slow movement. Nothing exotic required.
The recipe
- Two oscillators, a minor second apart. OSC A at the root, OSC B one semitone above. Both on sine. This is the specific interval, the tension that doesn't resolve because it isn't trying. A semitone sets the two fundamentals close enough to beat against each other inside one critical band, and the ear reads that roughness as tension.
- Both voices quiet, equal levels. OSC A at 60%, OSC B at 60%. Neither is the main note. They are a pair.
- A filter that muffles the clarity of the interval. the 24 dB/oct ladder lowpass, the steep kind, at 1400 Hz. Low enough to soften the beating between the two voices, high enough to keep them articulate.
- A long, slow LFO on OSC B's level only. LFO 1 as a sine at 0.03 Hz, depth 15%, on OSC B level. This makes the interval's tension breathe, the upper voice moves in and out of emphasis.
- A layer of air. Noise oscillator, pink, unfiltered, at 5%. The air is what makes the drone sit in a place rather than in abstract.
- Envelope that commits to the sound. Attack 2 seconds, release 4 seconds. The drone doesn't begin; it arrives.
Hold the note. The interval will do the work. You don't need a second chord. A held dissonance wants to be alone in its range, so give it a quiet bed and keep melodic parts well above it. For that bed, build Sunday Afternoon Pad and let it breathe softly beneath the drone.
Variations:
- Make OSC B a major second above (two semitones). The drone becomes less painful, more conventional.
- Remove the noise layer and the drone becomes clinical. Put it back and it's in a room again.
In the lineage of:
How to make a chiming, chorused pluck that rings into itself
Chiming Pluck. Each note rings into the next; the chord assembles itself in the room.
A bright plucked note, chorused past polite, let into a room large enough that the next note arrives before the last one has left. The source is a plucked string; the chorus, pushed well past subtle, is the instrument.
What you'll need: You'll need a synth with a physical-modeling or plucked-string oscillator (or a bright FM or triangle pluck as a stand-in), a short plucky amp envelope, heavy chorus, and reverb. Most synths cover all of these.
The recipe
- A bright plucked-string source. A Karplus-Strong plucked string (or a triangle source with a 1ms attack as a stand-in). Keep the body lowpass open enough that the upper partials read. Bell-bright, not warm. One voice, struck cleanly.
- Arpeggiate the chord. Walk a triad as four plucks (root, third, fifth, octave) staggered by ~280 ms so each note rings into the next. This is the chime arpeggio: the chord becomes a sequence, not a strum, each note ringing into the one after it.
- Heavy chorus, slow rate. Chorus depth 75%, rate 0.5–0.8 Hz, three-stage if available. Push past "subtle" until each pluck starts to swirl around itself. This is the move: the chorus is the instrument, not a polish.
- A long hall reverb behind it. Reverb decay 3.5 s, wet 40%. The chorus and reverb together are what put the pluck in a room large enough that the chord lingers between notes.
- Mix the jangle. A small high-pass at 100 Hz so the plucks don't fight the bass; 2:1 bus compression with slow attack so the arpeggio glues without flattening; a high-shelf cut of −2 dB at 8 kHz so the chorus stays warm rather than hissy.
Play the arpeggio slowly. The chorus and reverb hold the chord between each note. All that width and tail eats space, so commit to it and thin whatever else lives in the high mids. To give the chime something to ring over, lay Sunday Afternoon Pad underneath and let the arpeggio float on top of it.
Variations:
- Drop the chorus depth to 30% and the arpeggio becomes a generic jangle, the chime signature disappears the moment the chorus is gentle.
- Add a +12 pitch-shifted reverb return at 25% mix and the patch leans further into shimmer territory, the bells gaining ghosts an octave up.
In the lineage of:
- Cocteau Twins (Cherry-Coloured Funk)
- Slowdive (Catch the Breeze)
- The Smiths (This Charming Man)
How to make a plucked harp with physical modeling
Harp Pluck. Physical modeling stands in for the harp.
A handful of notes rolled like a real harp, though nothing here is a string. Physical modeling stands in for it; sampling isn’t the only path to real.
What you'll need: You'll need a synth with a physical-modeling or plucked-string oscillator, a lowpass filter, and reverb. Many synths include a string model.
The recipe
- A short noise burst into a tuned delay line. In Karplus-Strong synthesis, the source is a tiny burst of filtered noise; the resonator is a delay line tuned to the note pitch with a lowpass in the feedback path. Use a triangle oscillator with a 1ms attack as a stand-in if your synth doesn’t expose a KS engine directly. You'll know the delay line is tuned when a single strike rings like a plucked string, a clear pitch with a soft decaying tail rather than a click or a synth blip.
- A pluck envelope on the source. AMP ATK 1ms, DEC 800ms, SUS 0%. The note rings out for almost a second, then dies.
- Open the lowpass to brighten the body. Lowpass cutoff up around 3.5 kHz with the resonance kept flat. Opening the cutoff lets the harmonics that give a plucked string its wood-and-wire body through; a flat resonance keeps the tone musical, not whistly.
- Sequence a pentatonic figure. The pluck reveals itself in motion, play a pentatonic pattern at a moderate tempo. The patch is the source for melodic figures that feel acoustic.
- Place the harp in folk space. Gentle chorus, an analog-style ping-pong delay at dotted-eighth (the rhythm of the delay does some of the writing for you), and a plate reverb at 3 s decay. Mono compression with a slow attack so the pluck's attack stays clear but the tail breathes. Pan center, the stereo lives in the effects, not the source.
A folk-feeling line, made of nothing acoustic. It reads as delicate, so keep it high and a little off-center, placed with reverb rather than volume. When it wants a partner in the same fragile register, build Music Box Pluck and let it answer the harp between phrases.
Variations:
- Slow the decay to 200 ms and the patch becomes a wood block, sampling envelope shifts the perceived instrument.
- Add a second voice tuned a fifth up and the patch becomes a harp doubled at the interval, closer to Caribou than to Four Tet.
In the lineage of:
How to make a slow-bloom ambient pad, the swell envelope
Slow-Bloom Pad. Half of ambient borrows from this one envelope.
An envelope longer than memory. A reverb longer than the song. The slow-attack pad half of ambient is built on.
What you'll need: You'll need a synth with a unison control, an amp envelope with a very long attack and release, and a large reverb. Common building blocks.
The recipe
- Three saws, gently detuned. OSC A on a soft saw, unison 3 voices, detune ±4 cents. Just enough motion that the chord doesn’t feel digital, not so much that the patch swirls. At a few cents apart the voices beat slowly against one another, and that slow drift is what the ear reads as analog.
- A three-second attack. AMP ATK 3000ms, three full seconds. The note arrives so slowly you forget where it started. Long release (4 s) so the patch bleeds between chord changes.
- A twelve-second hall reverb. Reverb decay 8–12 s, wet 65%. The reverb is half the patch, without it, the slow attack is just a pad. With it, the pad is a room.
- Change the chord once a measure. Hold a chord for 8–16 beats. By the time the new chord arrives, the previous one is still finishing its release. Overlap is the point.
- Don't compress the bloom. The envelope IS the patch, compression flattens it. Instead, warm it with a +2 dB low shelf at 200 Hz, take −1 dB off a high shelf at 6 kHz so the saw never bites, and run the 12-second hall on a parallel bus with 200 ms predelay so the dry attack speaks before the cloud arrives.
Hold a chord. Change it when the room asks for it. This is background by design, so sit it low and wide under everything and let the long attack stay clear of transients. When you want a voice to move above this bed, scatter a Harp Pluck over the bloom, each note a bright point the pad swells up to meet.
Variations:
- Drop the attack to 500 ms and the patch becomes a slow-attack pad rather than a bloom, closer to ambient pop than to canonical ambient.
- Disable the chorus and the patch becomes a still pad, the canonical Eno register is in the slight chorus motion under the stillness.
In the lineage of:
How to make a string-machine section from twelve detuned voices
String Section. Twelve voices holding the same chord, none of them perfectly.
The way a string section sounds when you sit in the front row, twelve voices, each one slightly its own, breathing as a unit.
What you'll need: You'll need a synth with a unison (voice-stacking) control, a lowpass filter, an LFO for vibrato, and an amp envelope. All standard.
The recipe
- Twelve voices, wide section detune. OSC A on soft saw, unison 12 voices, detune spread ±14 cents, stereo width 0.85. The section is the sum of independent players. Real string-ensemble tuning variance sits in the ±10 to ±15 cent range, tighter than that reads as unison; wider than that reads as out of tune.
- Natural bow attack, committed release. AMP ATK 300ms, full sustain, REL 1.5 s. A real bow takes 200 to 400ms to engage the string; longer than that and the patch reads as synth pad rather than ensemble. The release is long enough to feel intentional, short enough to not smear into the next chord.
- Body-formant peak, not just a lowpass. A peaking EQ around 2.2 kHz, Q 1.3, +5 dB on top of the lowpass at 3 kHz. The peak approximates the violin/cello body cavity resonance, the small bump in the spectrum where real string bodies sing. That bump is the difference between "filtered saws" and "instrument."
- Late-onset vibrato, expressive width. Vibrato 4.6 Hz at ±12 cents, onset 700ms. Each note bows in flat, settles, then blooms into vibrato. The bow waits, the player adds it. In the final patch each voice has its own slightly-jittered vibrato so the section sounds like individual players, not synchronized synth strings.
- Bow noise, the scrape and the breath. A small noise generator, lowpassed at 8.5 kHz, 6% level, enveloped with the strings, plus a short bandpassed burst at note onset, the scrape of horsehair engaging the string before the tone arrives. Cut the bow noise and the patch goes synthetic instantly, a small parameter with outsized identity.
- A hall, not a smear. Gentle plate reverb at 2.5 s decay, 55% wet, concert-hall tight rather than ambient-wash long. Depth comes from the twelve independent voices; the room just places them in space. Chorus stays minimal, the player detune doing the work a Solina faked with chorus.
Hold a chord. Notice how little is doing the work, no chorus, no movement plugin, just twelve voices breathing together. Strings crowd the midrange fast, so ease a little around 400 Hz when a vocal is present and let the ensemble sit wide and back. That is how a real section sounds. To give the section a voice to answer, lay a Smooth Sax over the top, a single line leading where the ensemble holds the chord.
Variations:
- Push chorus to 60% and the patch immediately reads as Solina rather than real strings; chorus is the synth-machine giveaway.
- Drop the voice count to 4 and the section becomes a string quartet, intimate rather than grand.
- Slow the vibrato to 3.5 Hz at ±15 cents and the section ages thirty years, closer to a 1970s romantic film score.
- Bring the attack back up to 700ms and the bowed character disappears; the patch goes back to sounding like a synth pad pretending to be strings.
In the lineage of:
How to make a shoegaze wall-of-sound chord with no edges
Wall of Sound. A chord with no edges, arriving from everywhere at once.
Take the detuned-saw stack and keep stacking, more voices, wider, until you can no longer pick out a single one. Saturate it so the edges melt together, drown it in reverb, and the chord stops coming from the speakers and starts coming from the walls.
What you'll need: You'll need a synth with a unison (voice-stacking) control and a lowpass filter, plus saturation and a large reverb. Standard fare.
The recipe
- Start from the stack, then double it. OSC A on a bright saw, unison voices as high as the engine allows. Where the String Section kept twelve players distinct, here you want the opposite, enough voices that no single one is findable.
- Detune past the section, into the smear. Detune spread around 35 cents, stereo width nearly full. Past about twenty cents a stack stops sounding in tune and starts sounding like one wide breathing thing. That smear is the point, not a flaw to correct.
- Saturate until the voices fuse. Drive the whole stack into saturation, around 33 percent. Saturation is the glue, it rounds the separate detuned edges into one surface and fills the gaps between voices the way a wall of layered takes does. Clean, the stack is a chord; saturated, it is a wall.
- Soften the attack so it arrives, not strikes. AMP ATK near 700ms, long release. The wall does not start, it gathers. Slow enough that no note has an onset you can point to, fast enough that it is not the Slow-Bloom three-second bloom.
- Put it in a large room. Reverb decay 7 to 9 s, around 85 percent wet. At this depth the reverb is not an effect on the wall, it is part of the wall, an echo chamber the size of the room, and the reverb and the stack become one sound around the listener.
Hold one chord. Do not change it for a long time. The wall is not going anywhere, and neither, for these few minutes, are you. A wall like this is the whole bed, so build the track on top of it and high-pass it under the kick so the low end stays defined. Next, lay a String Section over the top so a few distinct players float above the smear and give the ear one moving line to hold.
Variations:
- Add a slow shared pitch drift, under 0.1 Hz and a few cents, and the wall breathes; leave it perfectly still and it becomes a held sample, beautiful but dead.
- Pull the saturation back to 10 percent and the voices separate again, the wall relaxing into a wide pad, a looser cousin of the String Section.
- Add an octave doubling underneath for a floor and the wall leans toward drone; raise the cutoff instead and it pushes forward into the room rather than wrapping around you.
In the lineage of:
How to make a vaporwave Macintosh chord, the major-key future sound
Macintosh Chord. The major-key chord that meant the future once.
The chord no one was paying attention to, finally the loudest thing in the room. Three detuned triangle voices voiced as a major seventh, slow and chorused, pitched down into vapor.
What you'll need: You'll need a synth with enough polyphony to hold a chord, a unison control, chorus, and reverb. All standard.
The recipe
- Three voices, major seventh voicing. OSC A at root, OSC B at major third, OSC C at major seventh. All on triangle. The chord is the patch, you hold one note, a major-seventh plays. You'll know the voicing is right when the top of the chord glows rather than bites, the major seventh shimmering just under the root instead of clashing with it.
- Slow chorus, very present. Chorus module, rate 0.5 Hz, depth 45%, feedback 25%. Heavier chorus than most patches allow, the mallsoft signature: a short modulated delay bends pitch the way tape wow does, which is why this much of it reads as warped tape. The chord widens and starts to swim, the pitch wobbling slowly side to side rather than sitting still.
- Mallsoft filter curve. the four-pole ladder lowpass, a steep rolloff, at 2200 Hz. The 24 dB per octave slope strips the brittle top harmonics fast while passing the band just under the cutoff, softening the surface without muffling the body. The tell is the top losing its glassy edge while the chord still reads as bright, soft on the surface yet not muffled underneath.
- A long reverb, bright tail. Plate reverb, 4-second decay, damping 30%, predelay 80ms. A bright tail opens out behind the chord and hangs for a few seconds, the room arriving a beat after the note so the attack still reads before the space swallows it.
- Pitch-shift down for the vapor effect. Fine tune the patch down around 50 cents, not an octave, just enough to feel slowed. This is the chopped-and-screwed inheritance in the vapor lineage.
- A soft attack, generous release. Attack 400ms, release 2 seconds. The chord should fade up instead of starting, with no point where it switches on, then linger for a couple of seconds after you let go.
One chord. Let it sustain for 30 seconds without playing a second. The register rewards patience. It is wide and slow by nature, so let it own the stereo field and keep everything else sparse around it. Into the space you left, a slow Crystal Piano arpeggio answers far back and drenched, one hammered line replying to the held chord from across the atrium.
Variations:
- Remove the pitch-down and the chord becomes merely a mall PA, the vapor character was entirely in that detune.
- Raise the chorus feedback to 50% and the chord becomes a tape loop of itself.
In the lineage of:
- Macintosh Plus (Floral Shoppe)
- Blank Banshee (Blank Banshee 0)
How to make a bright mallsoft mall marimba
Mall Marimba. Mallet-bright digital sound that carries past whoever's listening.
Mallet tones drifting across an empty food court, bright and going nowhere. The FM bell sound is the Windows 96 signature, sustained, never quite arriving.
What you'll need: You'll need a synth that can do FM with an envelope on the FM amount, a short percussive amp envelope, a lowpass filter, and reverb. Any FM synth covers it.
The recipe
- A two-operator FM pair, mallet ratio. OSC A sine carrier, OSC B sine modulator at 7:1 ratio. Modulation index around 0.9, enough metallic harmonics to read as marimba, not so much that it becomes a bell. The index sets how many sidebands the modulator spawns, so raising it slides the tone from struck wood toward struck metal.
- A sustained envelope. AMP ATK 8ms, DEC 800ms to 60% sustain, REL 1.2 s. The patch is held, this is the Windows 96 move that turns a pluck into a chord.
- Chorus and long reverb. Chorus depth 30%, rate 0.7 Hz; reverb decay 4.5 s, wet 55%. The mallet now floats, the mall PA reverberant register. Each strike lifts off and smears into the tail rather than landing on a surface, the note going nowhere and taking its time about it.
- Lowpass at 6 kHz. Gentle lowpass to remove the brittle top, keep the brightness, not the edge.
- Place it in the mall. Alternate notes panned ±40% (the marimba's natural stereo image as a percussion section would have it); a tape-style high-cut at 10 kHz so the mallets don't bite; the long hall reverb on a separate bus so the bell-pop survives but the chord floats above the floor of the song.
Hold a triad, the chord shimmers without the player adding anything. Bright mallets carry easily, so keep them quiet and high, a sparkle on top rather than a lead. The melody belongs underneath: a Y2K Glide Lead carries it while the marimba stays the high glint above.
Variations:
- Drop the sustain to 0% and the marimba becomes a pluck, the chord becomes ornamental rather than structural.
- Raise the modulator ratio to 11 and the patch becomes a tuned glass, further from marimba, closer to a bell.
In the lineage of:
- Windows 96 (Rituals)
- Macross 82-99 (Sailorwave)
How to make a crystalline mall piano with a long halo
Crystal Piano. Hammered tone with a halo that outlasts the note.
The Muzak version of a song you half-remember, played through speakers you can't see. An FM electric piano under a long, bright reverb halo.
What you'll need: You'll need a synth that can do FM, a short percussive amp envelope, and a long, bright reverb. Most FM-capable synths cover all of these.
The recipe
- A bell-FM patch, the canonical electric piano of the era. OSC A on sine, with OSC B frequency-modulating it at 45%. OSC B on sine, level at 0 (modulator only), tuned one octave above A. This is the FM electric piano that lived in every mall. You'll know the modulation depth is right when each note has a soft bell-like tine on the front, glassy but not metallic, the hammer in the tone without it turning into a chime.
- A short, percussive envelope. Attack 5ms, decay 800ms, sustain 30%, release 1000ms. The piano attack, then a long tail: a quick percussive front that drops fast to a low hum and rings out slowly, struck rather than bowed.
- A bright, slightly metallic filter setting. Filter 1 as lowpass at 4000 Hz, resonance at 10%. You want the FM sparkle preserved.
- The move that makes it mall: a very long bright reverb. Plate reverb, 5-second decay, damping at 25% (bright), predelay 60ms. The space is big and hard-surfaced. The tell is the halo outlasting the note, a bright tail hanging for several seconds after the key is gone, like the sound is reaching you from across a far-off atrium.
- Chorus after the reverb. Chorus module, rate 0.6 Hz, depth 20%. This is the pitched shimmer that says "speaker system", the stereo doubling that wasn't in the original signal.
- Slight pitch modulation. LFO 1 on OSC A fine tuning, sine at 0.5 Hz, ±3 cents. The PA speakers had slow pitch drift. The pitch sways gently in and out of true, barely there, enough to feel like aging speakers rather than something you would call out of tune.
Play a slow arpeggio in a major key. Resist the temptation to make it a full song. Drenched in reverb it wants space, so set it back in the mix and let the tail carry it instead of level. Beneath the arpeggio, a held Macintosh Chord makes a slow bed, so the piano falls like single drops over a chord that never moves.
Variations:
- Halve the reverb decay and the mall becomes an office lobby.
- Increase the FM depth to 70% and the piano becomes a bell, a different era of mallsoft.
In the lineage of:
- Macintosh Plus (Floral Shoppe)
- Windows 96 (Enchanted Instrumentals)
How to make a Y2K glide lead with a soft, sliding pitch
Y2K Glide Lead. Smooth, slid pitch, soft attack, pure 99/00 register.
A lead that was sure the future would be smooth, plastic, and chrome. A bright, wide stacked-saw glide lead, all Y2K optimism.
What you'll need: You'll need a synth with a unison control, a bandpass filter, a quick amp attack, chorus, and a short delay, plus portamento (glide) for the slides when you play it as a line. Most synths have all of these.
The recipe
- Bright detuned saw unison. OSC A on saw, 5 voices unison, detune 20 cents, stereo spread wide. The signature stacked-saw lead of the Y2K-futurism moment. The single voice thickens and spreads to the edges of the field, shimmering and wide, while still reading as one bright note.
- A highpass to keep the lead above the bass. Filter 1 as a 12 dB/oct highpass at 200 Hz. This is the move that makes the lead feel weightless. Clearing the low-frequency energy the ear reads as body leaves only the bright upper range to float.
- Bright bandpass on top. Filter 2 as Bandpass around 3000 Hz with moderate resonance. Serial filtering: Filter 1 into Filter 2. The bandpass is what gives the lead its specific "future" timbre.
- Chorus with visible motion. Chorus rate 1.5 Hz, depth 30%, feedback 20%. The chorus is fast enough to be perceived as motion, not just width.
- Short attack, moderate release. Attack 8ms, release 400ms. This is a lead, you hear the note's beginning clearly.
- Slight delay for space, not echo. Delay at 1/16 time, feedback 20%, mix 20%. The delay is a texture, not a repeat.
Fast riff in the upper-middle register. This lead rewards motion. Wide and bright, it takes the front, so thin the pad beneath it and give the glides room across the stereo field. That thinned bed is a held Macintosh Chord, slow and wide, so the lead glides over a chord that stays still.
Variations:
- Drop the bandpass filter and the lead becomes a generic detuned-saw lead; the Y2K character was in the specific bandpass placement.
- Slow the chorus rate to 0.3 Hz and the future slows down; the lead becomes early-2000s instead of late-1990s.
In the lineage of:
- Alice Deejay (Better Off Alone)
- Darude (Sandstorm)
How to make smooth Y2K lobby brass
Lobby Brass. Smooth, bright, and certain the future would arrive on schedule.
The smooth, bright brass that played under every corporate intro and hold-music loop, all confidence and good news. Two operators in a 1:1 ratio, a little feedback for rasp, and an envelope that flares the tone bright on the attack and settles as it holds, the way a real horn does.
What you'll need: You'll need a synth that can do FM, one sine modulating another, with an envelope on the FM amount, plus a lowpass filter and reverb. Most FM-capable synths cover all of these.
The recipe
- Two operators in a 1:1 ratio. OSC A as a sine carrier, OSC B modulating it, both at the same frequency. Integer ratios give harmonic, in-tune timbres, and 1:1 is the brass and organ register where every harmonic lines up over the fundamental. You'll know the ratio is right when the tone stays solidly in tune as it brightens, the harmonics stacking cleanly over the root with no clang or beating between them.
- Add FM amount and a little feedback for the bite. Modulation index around 55 percent, with a touch of feedback. This is where the tone stops being a flute and grows the buzzing edge of brass. Too little and it stays a sine, too much and it tears into laser and metal.
- Let the attack flare bright, then settle. Set the modulation to burst higher at the very start and fall back as the note holds, so the tone blares on the attack and mellows as it sustains, the way a horn blares as the player leans in. This is the difference between a static FM tone and brass that breathes.
- Keep the top bright. A gentle lowpass up around 4 kHz, just enough to take off the digital fizz and the laser sheen while leaving the brass glassy. The era wanted brass that sounded like a compact disc, not like a room.
- Chorus and a lobby reverb. Chorus around 22 percent for the CD-era stereo shimmer, then a medium bright reverb, 3 to 4 s, that places the brass in a carpeted corporate lobby rather than a concert hall. Present, optimistic, a little unreal. The brass widens and sits just off the speakers in a small bright room, close and forward, the tail short enough that it never feels like a hall.
Play a major triad, held, with conviction. This is the sound of being told everything is going to be fine. Brass sits in the upper mids, so let it lead and pull other midrange back while it plays. In the gaps, a Crystal Piano line answers far back and bright, a single voice replying each time the fanfare rests.
Variations:
- Pitch the whole patch down two or three semitones, the Macintosh Plus move, and the corporate optimism curdles into something wistful, the future remembered instead of promised.
- Drop the FM amount and the brass softens toward a flute or a soft lead; lower the carrier and it becomes a mellow horn rather than a fanfare.
- Raise the feedback and the brass grows a reedy, almost saxophone rasp, the smooth-jazz cousin of the same patch.
In the lineage of:
- Blank Banshee (Teen Pregnancy)
- James Ferraro (Linden Dollars)
- Saint Pepsi (Cherry Pepsi)
How to make a soft food-court sublevel bass
Sublevel Bass. The low end of music nobody chose to hear.
The low end of music no one chose to hear, smooth and fretless, sliding between notes under the hum of a food court. An FM bass at a 1:1 ratio, where a short burst of modulation on each attack gives the note a soft fingered mwah before the filter closes behind it.
What you'll need: You'll need a synth that can do FM with an envelope on the FM amount, plus a lowpass filter with a second envelope routed to its cutoff, and reverb. Standard on an FM-capable instrument.
The recipe
- An FM bass at a 1:1 ratio. OSC A on sine, OSC B modulating it at the same frequency. Keep the modulation amount low, around 15 percent, so the held tone stays round and warm rather than metallic. This is the smooth-jazz low end, the FM bass that lived in every Sega Genesis.
- A soft bark on the attack. A burst of extra modulation at the very start of each note, decaying over about 90ms. This is the soft pluck of the string, the bite before the body, the brief spike in modulation index throwing extra sidebands that brighten the attack alone. You'll know it lands when each note opens with a faint fingered mwah that has already faded before the body settles, and you've pushed too far when the front edge clacks instead of breathes.
- The fretless mwah. A filter envelope that starts open and closes over about 170ms, a negative envelope depth. The cutoff falling as the note begins is the mwah, the sound of a finger sliding on a fretless neck.
- A rounded amp envelope. Attack 10ms, decay down to a low sustain, release a touch over 200ms. The note arrives soft and leaves soft, nothing percussive about it.
- Barely any room. A whisper of chorus and a small, short reverb. This bass plays close to the floor, in a space with a low ceiling and carpet on everything. The tail should die almost as the note does, close and carpeted; if it ever opens into anything you would call a hall, the room is wrong.
Walk a slow line under an imaginary saxophone. Let each note round off before the next one starts. A round, sliding bass holds the low mids, so keep it mono and let the slides fill the space a kick would leave. That imaginary saxophone has a real partner in Lobby Brass: float it in soft chords above and let the bass walk underneath.
Variations:
- Raise the modulation amount and the bass sharpens toward a slap, brighter and more eighties.
- Open the filter and remove the envelope and the mwah disappears, leaving a plain round sub.
In the lineage of:
- Night Tempo (Ladies in the City)
- Yung Bae (Bae 3)
How to make a self-crossfading screensaver pad
Screensaver Pad. A chord that crossfades into itself while nobody watches.
A major-ninth chord left holding on an idle machine, crossfading into itself the way a screensaver dissolves one image into the next. A wavetable pad morphing between triangle and square, five voices detuned wide, a slow filter drift breathing underneath.
What you'll need: You'll need a synth with a wavetable oscillator, a unison control, a lowpass filter with an LFO on its cutoff, and chorus. Most synths have all of these.
The recipe
- A wavetable that morphs on its own. OSC A on a wavetable that runs from triangle to square. Set a very slow morph, around 0.08 Hz, so the timbre crossfades by itself while you hold a chord. Nobody is at the keyboard; the sound has to move on its own. You'll know the morph is slow enough when, holding one chord, you cannot point to the moment the tone changed, only that it has.
- A major-ninth voicing. Voice the chord as a major ninth: root, third, fifth, seventh, ninth. The wide, glassy, not-quite-resolved stack is the brief late-nineties optimism the patch is named for.
- Five voices, detuned wide. Unison five voices, detune around twelve cents, full stereo width. The slight detune sets the voices beating slowly against each other; the small spread keeps them distinct rather than fusing, so the result stays glassy and shimmering instead of thick.
- A slow filter drift. An LFO on the cutoff, very slow, around 0.06 Hz, with a small depth. The pad breathes without anyone touching it.
- Fade up, fade down. Attack around 1.2 s, release around 2.6 s. It wakes like a monitor coming on and sleeps like one going dark.
- Chorus and a long reverb. Heavy chorus, around 45 percent, and a long bright reverb. The shimmer belongs to the screen, not to the room. It should hang in the air long after the chord, wide and weightless; pull the reverb back and it suddenly belongs to your room again, which is the wrong room.
Hold one chord and walk away. Come back in a minute; it will have changed without you. This is set-and-forget background, so sit it low and wide and keep its slow crossfade clear of anything with a transient. A Macintosh Chord sits well in front of it, struck and short, answering the drifting pad behind.
Variations:
- Speed the morph toward 0.3 Hz and the pad shimmers instead of drifting, a busier and more anxious machine.
- Drop the ninth and it becomes a plain major triad, the same optimism with less future left in it.
In the lineage of:
- Windows 96 (Whispering Through the Aisles)
- 2814 (Birth of a New Day)
How to make a dial-up modem chord
Dial-up Chord. The sound of trying to get somewhere.
A two-voice square chord a fifth apart, its sustain cut away, gated into a stutter so each hit lands and falls before the next. Pitch and noise bursts snap at every attack, the chatter of a line trying to connect. Stop-start and searching, the sound of trying to get somewhere. Drop it where a track needs motion without a melody: a rhythmic bed under a vocal, or the pulse between the kick and the lead.
What you'll need: You'll need a synth with a square oscillator, a way to play two voices a fifth apart, a bandpass filter, a very short amp envelope, and a tremolo or gate you can run fast for the stutter. Most synths have all of these.
The recipe
- A stack of detuned squares. OSC A on square at root, OSC B on square at a perfect fifth. Both with slight detune. The square carries the "digital signal" quality: a hard, hollow ring with a faint waver between the two voices, brittle rather than warm.
- Abrupt pitch bursts at the attack. Modulation envelope (ENV 2) on oscillator fine tuning, very fast attack, fast decay. Depth: ±50 cents. Each note starts with a little pitch jump, the modem's tone negotiation.
- A short, filtered noise burst per note. Noise oscillator enabled, one-shot mode, very short envelope (attack 2ms, decay 80ms). This is the initial handshake burst.
- Bandpass filter, narrow. Filter 1 as a 12 dB/oct bandpass at 1500 Hz, resonance 40%. This is the phone-line bandwidth, you're simulating the actual frequency range of a telephone. You'll know the band is right when the lows and the air are both gone and what is left sounds like it is coming through a handset; open the bandpass wider and it stops sounding like a phone.
- No reverb. The dial-up happened in a computer, inside a room, without ambience. Dryness is part of the period. Each hit stops dead the instant it ends, no tail at all, as if it happened inside a sealed beige box.
- Stuttered rhythm. Use an arpeggiator or clip with fast 1/32-note pattern, gate at 50%. This is the sound of data packets, not a held chord, a pulsing one.
Play a chord stab and let the dotted delay carry it across the bar. The space between the hits is the point. Plucked and delayed, it wants air, so leave the off-beats open and let the dotted delay do the rhythmic filling. A Fax Machine Bass belongs in those gaps the chord leaves open, its short pulses locking under the stutter.
Variations:
- Lengthen the decay to a half-second and the pluck softens toward a pad, the same chord, less urgent.
- Push the delay feedback past 50% and the hits smear into a wash; back it off and each one stays separate.
In the lineage of:
How to make a worn VHS pad with warm tape hiss
Worn VHS Pad. Warm sounds from a tape that's been played too often.
A pad that refuses to stay whole. It starts smooth, then the grains stop agreeing, each at a slightly different pitch, and the heavy tape saturation grinds a rough metallic edge onto them, until the chord sounds recorded onto a tape that was already worn. Damaged on purpose. Reach for it when ambient needs texture that feels degraded and a little haunted, not clean.
What you'll need: You'll need a synth with a unison control you can detune heavily, heavy saturation or tape-style drive for the metallic edge, a lowpass filter, and reverb. Nothing exotic required.
The recipe
- A warm, heavily saturated pad. OSC A on saw, unison 3 voices, detune 8 cents. Tape-style saturation at 40%. The drive is heavy, you want the "VHS warmth" to be immediately audible.
- A significant pink noise layer. Noise oscillator, pink, level at 15%, much higher than most of our pads. The hiss is a co-equal voice in this patch.
- Pitch drift, unmistakable. LFO 1 on OSC A fine tuning, sine at 0.4 Hz, depth ±10 cents. The tape's transport is not stable. You'll know it reads as tape when the pitch will not sit still, swimming gently flat and sharp; push the depth far past this and it tips from worn into seasick.
- Heavy high-end rolloff. the 24 dB/oct ladder lowpass, steep on the highs, at 2800 Hz. Then additional rolloff via a shelving EQ if available. The VHS top-end is missing.
- A slight bass boost. Low shelf EQ at 200 Hz, +3dB. VHS tape had a specific bottom-end warmth from its frequency response. The tell is the bottom thickening into something warm and close, fuller without turning boomy or muddy.
- Gentle chorus, slow. Chorus rate 0.2 Hz, depth 20%. This recreates the phasing of tape head alignment drift: a chorus blends in a short, slowly modulated delayed copy, and its moving comb notches are the sweep you hear.
Hold a simple chord and let it come apart. The fraying is the music, not a flaw in it. It spreads across the spectrum, so sit it low under the track and let the instability read as texture, not pitch. A Warped Lead drifts well over this bed, its wow answering the pad's, two unsteady tapes leaning on each other.
Variations:
- Push the grain pitch-randomization past 25% and the voices scatter further. The pad stops being a chord and becomes a cloud.
- Back the saturation off below 20% and the damage softens to a sheen; push it past 50% and the body collapses into harshness.
In the lineage of:
How to make an AM-radio pad where signal and static blur
AM Radio Pad. Signal and interference have stopped arguing.
Somewhere, a station is playing. The atmosphere has opinions about which one you hear. A noise-based wash, tuned like a dial drifting between stations.
What you'll need: You'll need a synth with a noise source, an LFO routed to the amp for the radio tremolo, a lowpass filter, and reverb. A basic synth covers it.
The recipe
- Bandpass-filtered noise, wide resonance. Noise oscillator, pink, filtered through a 12 dB/oct bandpass at 2000 Hz, resonance 30%. This is the AM bandwidth, narrow, present, low-fi.
- A distant melodic ghost, the station itself. OSC A on triangle, played in the mid-register, level at 30%. Heavy lowpass on top, the station is buried in the noise.
- Amplitude modulation between the noise and the melody. LFO 1 on noise level, sine at 0.3 Hz, depth 40%. The signal comes and goes; the interference is periodic.
- A slow pitch wobble on the melodic ghost. LFO 2 on OSC A fine tuning, random shape at 0.2 Hz, depth ±8 cents. The station is drifting out of tune, atmospheric refraction.
- No reverb. AM radio doesn't have reverb; it has signal. The dryness is the register. You'll know it sits right when it stays flat and forward with no tail behind it, present rather than distant, the way a small speaker pushes sound straight at you.
- A very long envelope. Attack 1 second, release 3 seconds. The station tunes in; it doesn't start.
A single held chord. Let the interference be most of what you hear. It is mostly noise and air, so place it as atmosphere around the track, never where it has to carry a note. A Worn VHS Pad beneath it gives the static a warm body to wash across.
Variations:
- Narrow the bandpass resonance to 80% and the station becomes clearer, you can almost make out the melody.
- Cut LFO 1 and the interference stops cycling; the wash becomes static.
In the lineage of:
How to make a fax-machine bass texture
Fax Machine Bass. The specific texture of a machine that doesn't exist anymore.
The sound of a 1996 office at 5:47 PM, transposed to bass. A sharp, percussive FM bass with a digital-artifact bite at the attack.
What you'll need: You'll need a synth that can do FM with an envelope on the FM amount for the attack bite, a short percussive amp envelope, and a lowpass filter. Any FM-capable synth covers it.
The recipe
- An FM bass, sharp and digital. OSC A on sine, with OSC B frequency-modulating it at 50%. OSC B on sine at 2:1 ratio (one octave above). The FM depth is high, so the bass takes on a metallic bite: the high modulation index throws dense high-order sidebands, and those upper harmonics are the bite.
- A digital-artifact burst at the attack. ENV 2 modulating the FM amount: attack 0ms, decay 40ms, sustain 0%. The first 40ms of each note is a screaming FM burst. Then it settles.
- Phase-modulated decay. A second FM routing, a phase-distortion stage with self-feedback at 15%. This gives the decay a specific phase-warping character. You'll know it is working when the decay no longer falls cleanly but curdles, a faint metallic grain riding it down; push the feedback higher and that grain tips toward a buzz on the edge of breaking up.
- Keep the bass narrow. Unison off or at 1 voice. This is a mono, point-source sound.
- Filter stays mostly open. the four-pole ladder lowpass, four poles of steep rolloff, at 2000 Hz, resonance 15%. You want to hear the FM clearly.
- Percussive envelope, tight. Attack 0ms, decay 200ms, sustain 20%, release 150ms. The bass is a punch, not a hold.
Short bass stabs in a pattern. This sound doesn't sustain; it pulses. Percussive and bright, it locks to the rhythm, so keep it tight and mono and let the attack cut through the low end. A Dial-up Chord stuttering above it makes the pairing complete, the two machines pulsing the same office air together.
Variations:
- Soften the FM env attack to 20ms and the bass loses its digital "scream"; it becomes merely metallic.
- Raise the phase-distortion feedback to 40% and the decay becomes unstable, on the edge of self-oscillation.
In the lineage of:
How to make a tape-warped lead with pitch drift
Warped Lead. Pitch drift turning a melody into a memory.
The lead used to be clear. The tape has an opinion about that now. A triangle lead smeared by tape wow and flutter, melody bending into memory.
What you'll need: You'll need a synth with a unison control, a lowpass filter, and an LFO (two if you can) routed to pitch for the tape warp. Most synths have all of these.
The recipe
- A triangle-based lead with unison. OSC A on triangle, unison 3 voices, detune 10 cents. A triangle survives degradation because its energy sits near the fundamental with only sparse, fast-rolling-off harmonics, so the lowpass and tape losses to come barely touch it. The three voices should read as one soft, slightly thickened tone, a faint chorused shimmer at the edges and no audible beating.
- Heavy pitch-drift modulation. LFO 1 on OSC A fine tuning, smooth random (or slow S&H) at 0.6 Hz, depth ±15 cents. This is the "warp", the wow of a stretched tape.
- A second, slower pitch drift. LFO 2 on OSC A fine tuning, smooth sine at 0.1 Hz, depth ±6 cents. Two drift rates compounded, fast flutter plus slow wow. You'll know it is set when you hear two layers of unsteadiness, a quick quaver riding on a slower tide that takes several seconds to swing back; you've gone too far when any one note bends far enough that you notice the pitch leaving and returning.
- Tape saturation. Tape-style saturation at 30%. Warmth and compression: the soft-clipping rounds the loudest peaks down and folds low-order harmonics over the fundamental. The loud moments should stop poking out, the whole tone gluing and warming, the top edge softened rather than gaining grit you can pick out as fuzz.
- Filter softened. the 24 dB/oct ladder lowpass, steep enough to round the top off, at 2800 Hz. Gentle rolloff; keep the lead audible but tired.
- Slightly longer envelope than a normal lead. Attack 30ms, release 800ms. The notes don't end cleanly; the tape carries them slightly.
A slow melodic line. Don't play fast; the tape will distort the attacks and the line will become incoherent. Smeared and unstable, it works best alone in the midrange, so thin everything around it and let the wow be the only motion. Underneath it, a Worn VHS Pad makes a room for the lead to drift through, its hiss the air around the line.
Variations:
- Increase LFO 1 depth to ±30 cents and the tape is heavily damaged, the notes barely cohere.
- Remove LFO 2 and the lead becomes merely flanged; the compound drift is what creates "tape."
In the lineage of:
How to make a granular guitar cloud, melody without the guitar
Guitar Cloud. Melodic shape preserved; instrument identity lost.
A guitar dissolved into grains until only its shape drifts past. The source is barely identifiable; the melody is the only thing left.
What you'll need: This build asks more of your synth than most: a granular oscillator with control over grain density and length, grain position-randomization, and a long reverb. Not every synth has a granular mode. The audio under each step shows where it's headed.
The recipe
- Load a melodic instrumental source. Open the SAMPLE or GRANULAR oscillator. Load a guitar phrase, a piano figure, or any sustained melodic source. The melodic content is what survives the granular processing. At this stage the source still plays back close to normally, the instrument plainly recognizable, before any of the dissolving begins.
- Short grains, dense schedule. GRAIN LENGTH 50 ms, GRAIN DENSITY 15 Hz. Short enough that individual grains aren't identifiable; dense enough that the texture reads as sustained.
- A whisper of pitch randomization. Pitch jitter ±5 cents per grain. Tiny, the melody must stay intact, just enough to dissolve the source's timbral identity. Nudging each grain's pitch independently smears the fixed harmonic and formant relationships the ear uses to name an instrument.
- Heavy hall reverb. Reverb decay 6 s, wet 70%. The granular processing puts the source in a cloud; the reverb puts the cloud in a cathedral. You'll know it is right when each note keeps ringing for several seconds after you'd expect it to stop and the spaces between notes fill in with a smeared wash; pull the wet back if the dry sound disappears entirely under the tail.
- Prep the cloud. A small high-shelf cut on the source guitar before granular processing (granular adds its own brightness, you don't want it doubled); the cloud through hall reverb at 6+ seconds on a parallel bus; a faint pink-noise bed at −30 dB underneath. The cloud sits inside the noise rather than over it.
Let the melody play under the cloud. After thirty seconds you've forgotten the source was a guitar. A drifting cloud fills space without a clear pitch, so use it as a bed behind the track and let the melody float on top. Over the cloud, an Underwater Lead gives you the one moving line above the haze, dark and filtered enough to belong to the murk.
Variations:
- Reduce the pitch jitter to zero and the patch becomes a granular sustain rather than a cloud, closer to a freeze than a dissolution.
- Increase the grain length to 200 ms and the source identity returns, the lesson disappears when the grains are long enough to recognize.
In the lineage of:
How to make a saturated cassette tape bass with hiss
Forgotten Tape Bass. A saturated sub with tape hiss the same volume.
It was probably someone else's. There's no label on it. The first thirty seconds are mostly hiss before the tape kicks in. A saturated sub bass with the tape hiss mixed just as loud.
What you'll need: You'll need a synth with a sub or second oscillator, a lowpass filter, a noise layer for the tape, and an LFO routed to pitch for the wobble. Standard on most instruments.
The recipe
- Single triangle voice, low. OSC A on triangle at C2. One voice, no detune, this isn't a chord patch. The triangle gives you body without the brightness of a saw.
- A sub layer, sine, octave down. Enable the SUB oscillator on sine, minus one octave, level at 40%. The sub is what makes the recording feel old, analog tape compresses the low end into something that sits below the music. A new weight should settle under the triangle without becoming a separate note; if you can pick the sub out as its own pitch, it is too loud, ease it back until it only fills the floor.
- Tape pitch drift, wow and flutter. LFO 1 on pitch at 0.18 Hz, depth 8 cents. The slow wow that says "this tape has been in a hot car." Not enough to sound out-of-tune; just enough to feel found.
- Saturation past comfort. Insert a distortion/drive stage on OSC A at 35-45%. The triangle gains warm harmonics, then character; pull the drive up until the sound starts to bite, then back off slightly. This is the Tascam 4-track preamp running hot.
- Pink noise at recordable volume. Enable the NOISE oscillator at pink, level around 10%, louder than a conventional patch, this is intentional. The hiss should sit just under the source like it does on a real cassette: audible, structural, not a mistake.
- Lowpass everything together. Filter 1 as a 12 dB/oct lowpass at 950 Hz. Roll off the high end the way magnetic tape would, not a wall, just gentle attenuation so the hiss reads as warm rather than crisp.
Hold a low note. The bass sits, the tape drifts, the hiss holds the room. This is what bass sounded like before there were specifications. It is meant to be felt low and warm, so keep it mono at the bottom and let the hiss sit just under everything as air. A Warped Lead in the midrange above it shares the same tired tape while the bass holds the floor.
Variations:
- Bypass the saturation and the patch becomes clean tape, closer to a home demo than a destination.
- Increase the noise level to 15% and the patch becomes the recording itself, the bass is barely there, the hiss is the song.
In the lineage of:
How to make a cloud-rap pad shaped like a vowel
Cloud Pad. A pad with the silhouette of a vowel.
A voice that was auto-tuned before anyone sang into it. A formant-filtered pad, pitch-snapped to a scale and drowned in reverb.
What you'll need: You'll need a synth with a unison control, a lowpass filter with an LFO on its cutoff, enough polyphony to hold a chord, and a large reverb. Most synths have all of these.
The recipe
- A formant-filtered pad. OSC A on a warm triangle, OSC B on triangle one octave up at 50%. Filter 1 as a formant (vowel) filter. The formant is what creates the vocal-adjacent quality, parking resonant peaks at the frequencies a human vocal tract produces so the ear reads a throat behind the tone. You'll know the formant is doing its job when the pad reads as if it is holding a single open vowel, a throat behind the synth, rather than a flat filtered tone.
- Slow formant sweep. LFO 1 on Filter 1 cutoff (which controls vowel selection in a formant filter), slow sine at 0.15 Hz. The pad moves between vowels, "ah" to "oh" to "ee", without committing to any.
- Quantized-sounding pitch. This is where the move gets specific: set the patch to snap pitches to the major scale (via Key/Scale). Play any note; it locks to the scale. Auto-tune's hard-quantize quality, the tell being that forcing pitch onto exact scale steps with no glide strips the continuous micro-variation a real voice never fully loses.
- Wide stereo via unison. Unison 3 voices, detune 10 cents, width 100%. The pad spreads to the far edges of the stereo field, drifting slowly between the sides; checked in mono it should still hold together and not hollow out.
- Heavy wet reverb. A hall reverb, 5-second decay, damping 40%, with internal chorus. The cloud in cloud-rap is largely reverb. The close, dry pad recedes into a long wet tail that blurs each note into the next, distant and wide rather than near; pull the wet back if the pad loses its edges entirely and turns to fog.
- Soft, patient envelope. Attack 400ms, release 2 seconds. Each note arrives gently and recedes.
Play a slow melody in the upper register, simple intervals. The scale-quantize does the work. Washed in reverb, it sits back and wide, so keep it under the vocal and let the long tail be the space. A Ghost Melody threaded across the top answers the pad with a thin figure from inside the same wash.
Variations:
- Turn off the scale-quantize and the voice-ness goes away; the pad becomes conventional.
- Raise the LFO rate to 0.5 Hz and the formant sweeps visibly, the voice is trying to speak a word.
In the lineage of:
- Bladee (Working on Dying)
- Ecco2k (E)
How to make a dark, near basement sub-bass
Basement Bass. Held sub, dark and near, the floor of a room with no ceiling.
You can hear the bass because it is travelling through the floor. You can't hear anything else. A detuned-saw sub, lowpassed to drywall thickness and saturated for weight.
What you'll need: You'll need a synth with a sub or detuned-saw oscillator, a lowpass filter you can close down hard, and plenty of saturation for weight. No specialized gear needed.
The recipe
- Detuned unison bass, saw. OSC A on saw, unison 3 voices, detune 15 cents, Linear mode. Played around G1-C2. What you get is a thick, restless low buzz, fuller than a single saw but not yet wide or dark, with a slow churn as the slightly mistuned voices beat against one another at their difference frequency.
- Heavy lowpass, drywall thickness. the four-pole ladder lowpass, steep and dark, at 400 Hz, resonance at 20%. Almost everything above 400 Hz is missing. This is the move that makes the bass sound like it is somewhere else: walls and floors swallow highs far more than lows, so shedding the top is exactly what the ear reads as sound arriving through a barrier.
- Narrow stereo. Unison width 40%. The bass is coming from a single location, another room.
- Saturated, weighty. Tube-style saturation at 35%. The saturation compensates for the high-frequency loss by adding harmonic body. You'll know it is right when the bass gains weight and presence without getting any brighter, as if it filled out from the inside; back off if it starts to buzz or rasp instead of thicken.
- Slow amplitude modulation. LFO 1 on amp level, sine at 0.4 Hz, depth 10%. The basement activity is breathing, the bass is louder when the door is closer to open.
- Percussive-but-not-too-sharp envelope. Attack 20ms, decay 400ms, sustain 50%, release 300ms. Each note lands with a soft, almost-immediate onset that settles to a steady held body, firm enough to feel deliberate but with no sharp click at the front.
A simple bass pattern, slow tempo. The pattern is the song; there is no other instrument audible from where you are. It is meant to be felt more than heard, so keep it mono and low and let it own the bottom with little above it. The same pattern rebuilt as a Cushioned Bass gives a warmer, softer floor, the choice being which room you want to be standing in.
Variations:
- Raise the cutoff to 1200 Hz and you're now in the basement rather than above it.
- Remove the tube saturation and the bass becomes merely a lowpass saw, the drywall is gone.
In the lineage of:
- Salem (King Night)
- Varg2™ (Nordic Flora Series)
How to make an 808 bass with a vowel in it
808 Bass. A sub-bass with a vowel in it.
A sub with a mouth. It falls into the root and opens like a vowel, then eases shut, the bass saying a syllable where another track would put a note. Built for half-time and dub, where weight does the work a melody might.
What you'll need: You'll need a synth with a sine or triangle oscillator, a pitch envelope you can route to oscillator tune, a resonant lowpass with its own envelope, and saturation. Most synths have all of these.
The recipe
- A sine or triangle sub, dropped low. OSC A a sine or triangle, tuned down to E1. A second voice fine-tuned a few cents off at low level keeps the note from sitting perfectly still. This is the body.
- The pitch drop, the wo. An envelope on oscillator tune, amount +12 semitones, attack 0, decay about 130 ms. The note starts an octave high and falls to the root in a tenth of a second. That fall is the wo.
- A resonant lowpass, the vowel. Route both voices through a lowpass with high resonance. Drag a filter envelope to cutoff: a fast snap open to around 950 Hz, then an ease back down through the decay. The cutoff opening and closing is the vowel, the sharp resonant peak moving like a single vocal formant, one vowel sliding into another.
- Saturation, so it carries. Tube-style saturation after the filter, driven up then backed off. The resonant peak grows harmonics, and those harmonics are what carry the sound on small speakers where the fundamental alone would vanish.
- Soft attack, long hang. Amp envelope: attack around 8 ms so it blooms rather than clicks, low sustain, release near 1.1 s so the note hangs after the vowel has closed.
It sits naturally in half-time, dub-leaning electronic music, where its weight and slow vowel do the work a melody might do elsewhere. Keep it low and let it own the bottom. Beneath it, a Garage Sub Drone holds the floor so it never lifts between hits while this bass speaks its syllable on top.
Variations:
- Shorten the pitch decay toward 60 ms and the wo turns into a sharper, trap-style drop; stretch it past 200 ms and it sags, drunker and dubbier.
- Add a slow LFO of a few cents on master pitch and a gentle high-frequency roll-off for the tape-worn, Modern Love flavor.
In the lineage of:
- Andy Stott (On Oath)
- Mssingno (XE2)
How to make a distorted, darkened cowbell
Distorted Cowbell. The most ridiculed sound in the box, dragged somewhere dark.
The cowbell, the most mocked preset there is, and somehow the single most recognizable sound in phonk. Two square tones a fifth apart make the metallic clonk no single tone has; a hard short envelope makes it a hit; and the distortion it inherited from Memphis rap gives it teeth.
The recipe
- Two tones, a fifth apart. A cowbell is two square tones tuned a fifth apart, near 540 and 800 Hz. Summed, their clash is the metallic clonk; neither tone alone has it.
- Band-limit to the clonk. A bandpass around 820 Hz, fairly narrow, keeps the metallic body and sheds the boom below and the fizz above. The cowbell lives in that one band; sweep the center until the clank is at its most hollow, then leave it.
- A hard, short hit. Fast attack, a short decay of about half a second, almost no sustain. The cowbell is struck and gone, not held. You'll know it is short enough when it lands as a hit you could place on a grid, not a tone that lingers into the next one.
- Distort it. Drive it hard into saturation until the clonk grows a buzzing edge and the tail starts to spit. That distortion is what carries it from novelty to menace. Push until the metal sounds angry, then back off the instant the pitch starts dissolving into fuzz.
- Dry and dark. Keep it close, a short room at most. The distortion is the space; let the hit stay in your face.
Loop it as a pattern and it stops being a joke: a dark, distorted clonk driving a track from underneath, the cowbell finally taken seriously. Drop it an octave under a half-time beat and let it answer the kick on the offbeats; that call and response is the whole phonk engine.
Variations:
- Clean electro: drop the distortion and brighten the band for the classic 808 cowbell.
- Industrial: more drive, a tighter band, a faster pattern.
- Dub: one hit, a long delay, let the repeats decay into the dark.
In the lineage of:
- Three 6 Mafia (Mystic Stylez)
- DVRST (Close Eyes)
- Kordhell (Murder In My Mind)
How to make a buried, over-saturated Reese bass
Buried Reese. A reese run through more saturation than it asked for.
A low growl pushed so far down it seems to surface through mud. The reese is two detuned saws beating against each other; the character is what the saturation does next.
What you'll need: You'll need a synth with a unison control, a sub or second oscillator, a lowpass filter, and plenty of saturation. No specialized gear needed.
The recipe
- Two detuned saws, slow beat. OSC A and OSC B both on saw, detuned 11 cents apart. The interference pattern between the two voices is the reese, slow phase cancellation creates the rolling growl. Aim for a growl that rolls at a slow steady rate rather than sitting still; too few cents and it locks into one dead tone, too many and the beat speeds into a buzz.
- A sub-octave drone underneath. Sub oscillator at one octave below the saws, level 35%. The sub is what gives the patch weight; the saws give it growl. The bottom should drop and steady when it comes in; if the growl thins out instead, the sub is buried and wants more level.
- Soft saturation, past comfort. Waveshaper at 55% drive, past where it feels safe. The saturation thickens the beat pattern and adds upper harmonics the source doesn't have. The tell that you have enough is a grainy upper edge riding on top of the growl; you have gone too far when the beat pattern smears into a flat wall of fuzz.
- Lowpass at 600 Hz, slight resonance. Tight lowpass to keep the patch in the bass register, with a touch of resonance to highlight the cutoff frequency. The buzz pulls back into the low mids with a faint hollow ring at the cutoff; if it starts to whistle, the resonance is too high.
- Parallel-compress the growl. Parallel compression at 8:1 on a copy of the patch (mixed in at 30%), the dry path uncompressed, both summed, sidechained to the kick. The reese breathes with the floor without losing weight, the parallel chain catching the saturation transients without flattening the saw beat. You'll know the blend is right when the growl ducks and recovers with each kick yet the low weight never drops out; if the saw beat goes lifeless, the parallel level is up too high.
Hold a low note, the beat pattern between the saws is the song. A reese fills the low mids, so keep it mono down low and carve room above for anything that has to be heard over it. A tighter Basement Bass beneath it keeps the sub anchored while the reese does its growling in the low mids.
Variations:
- Disable the sub layer and the patch becomes a mid-bass growl, less weight, more bite.
- Drop the saturation drive to 20% and the patch becomes a clean reese, the lesson disappears when the saturation is gentle.
In the lineage of:
How to make a UK-garage sub drone that drops in pitch as it fades
Garage Sub Drone. Held sub, dropping in pitch as it fades.
A low tone you feel in the floor more than hear, sliding downward as it goes. The bass is the held tone: a single sine sub with a downward pitch envelope, no unison, no filter motion. The click that arrives with it is a different patch entirely.
What you'll need: You'll need a synth with a sub or second oscillator, a lowpass filter, an amp envelope, and a little saturation. Most cover all of these.
The recipe
- A sine at the song’s root. Sub oscillator on sine at G1, 49 Hz. One voice, no unison. This is the entire bass, no harmonic support, no filter movement, just one tone. It is clean when nothing flickers or beats; a single sine has no internal motion, so any wobble means a second voice crept in.
- Slow attack, long sustain, no decay. AMP ATK 50ms, SUS 100%, no decay envelope on amp. The bass holds across the bar; the amp doesn't move.
- Pitch envelope only. ENV 2 → OSC A coarse tuning, ATK 0ms, DEC 800ms, SUS 0%, ramp from +1 semitone to 0. The pitch drops one semitone across the held note while the amp stays still.
- Lowpass at 200 Hz. Steep lowpass to keep the patch a pure sub, anything above 200 Hz is not the bass.
- Mix for the rig and the laptop. One move at a time. Parallel tube saturation near 30 percent on the wet copy grows a harmonic above the fundamental, so the bass survives on a phone speaker where the pure sub would vanish. A high-pass near 30 Hz keeps subsonic rumble from eating headroom, and the opening click stays on its own channel: bass is one patch, click another. You'll know the saturation is right when the low note is audible on laptop speakers but still feels like sub on a system.
One held note, the downward slide its only motion. Keep it alone at the bottom, mono and uncluttered, and let nothing else live down there with it. The click belongs to a separate patch, a short filtered-noise burst at the note onset, and the two together make the full Archangel-style bass and hit.
Variations:
- Bypass the pitch envelope and the patch becomes a static sub, Bowery Electric territory rather than Burial.
- Add a parallel noise burst (separate patch) at note onset and you have the full Archangel bass + click arrangement.
In the lineage of:
How to make a ghostly, dissolving melody you can almost name
Ghost Melody. A tune you can almost name, dissolving as you reach for it.
A melody you can almost name, granulated until only its shape drifts past. The pitch is still there; the voice that carried it is gone. A granular cloud built from a single sustained source, its grains scattered just enough to dissolve the timbre while the line survives.
What you'll need: This build asks more of your synth than most: a granular oscillator with grain position-randomization, a noise layer, and a long reverb to blur it all together. Not every synth has a granular mode. The audio under each step shows where it's headed.
The recipe
- Load a sustained melodic source. Open the granular oscillator and load something with a clear pitch and a slow shape: a held vocal note, a string, a simple tone you played in. The melodic content is the only thing that will survive. You'll know the source is right when you can still hum the pitch back after one pass; if there is no clear note to hold onto, nothing downstream will carry the melody.
- Dense grains, medium length. Grain length around 120ms, density around 32 Hz. Dense enough to read as one continuous tone, long enough to keep a trace of the source color.
- Scatter the read position. Randomize the grain position around twenty percent. The grains stop agreeing on where in the source they came from, and the timbre smears into fog while the pitch holds steady. The voice that carried the note dissolves while the note itself stays put; push the scatter too far and the pitch starts to wander and break apart.
- A breath of air underneath. A whisper of filtered noise under the cloud, very quiet. This is the tape hiss, the air of a recording too old to be clean. The test: you notice the air only once it is gone; if you can pick the hiss out as a separate layer, it is too loud.
- A space too large for it. Reverb decay around 8 to 9 s, very wet. The cloud is put in a room far bigger than the melody, and the tune arrives already a memory. The line should arrive already fading, more after-image than note; if the dry cloud still reads out front, the wet is too low for the room to swallow it.
Play a slow, simple line and do not crowd it. The ghost needs room to fade. Faint and drifting, it asks for emptiness, so place it alone with reverb and keep everything else well clear of it. When you want the air around it to thicken, drift a Guitar Cloud far behind the line, quieter still, so the melody fades into a haze rather than into nothing.
Variations:
- Lengthen the grains toward 250ms and the source begins to identify itself again, less ghost, more echo.
- Push the position scatter past forty percent and the melody dissolves entirely into texture.
In the lineage of:
How to make a PWM square-wave bass
PWM Bass. A square-wave bass with pulse-width modulation as the only character.
A square-wave bass that seems alive, widening and narrowing in place. When the pulse-width is the source's animation, the bass breathes without anything else having to move.
What you'll need: You'll need a synth with a pulse or square oscillator whose width you can modulate with an LFO, a lowpass filter, and a compressor. Most synths have all of these.
The recipe
- A single square wave, low. OSC A on square at E2 (82 Hz). One voice, mono. The square is the carrier of the breathing pulse-width. The raw square should sit dead still, hollow and reedy with no motion at all; any wobble now means a modulator is already on.
- PWM via LFO. LFO 1 on OSC A pulse-width, sine at 0.6 Hz, depth ±40%. The duty cycle now sweeps narrow to wide and back, reshuffling which harmonics are present, and that shifting spectrum is the breath itself. The timbre should swell from thin and nasal to round and full and back, roughly one breath every couple of seconds; speed the rate up and it stops breathing and starts to flutter.
- No filter movement. Lowpass at 1.8 kHz, static. The lesson is PWM, not filter sweep, keep the cutoff still so the only motion is the pulse-width. The brightness should hold flat while only the body of the tone moves; if you hear the top end opening and closing, the filter is sweeping and stealing the lesson.
- Soft compression. Compressor across the output with slow attack (50 ms), gentle ratio. Tames the dynamic shifts from the PWM without flattening the character. The level steadies from note to note while the breathing still moves underneath; if the pulse-width motion goes flat, the ratio is too hard.
- Mix for the floor. High-pass at 30 Hz for sub rumble, soft 4:1 compression keyed off the kick so the bass ducks 2–3 dB on every quarter note, and a low-shelf of +1.5 dB at 80 Hz so the PWM reads on small speakers. Set right, the bass keeps body on a laptop yet ducks cleanly under each kick on a system; if the low end disappears on small speakers, the shelf is doing too little.
Play a sustained note, the bass breathes on its own. It moves on its own, so keep it mono and low and let the pulse-width animation stand in for other low-end movement. For the stiller stretches, swap in a Cushioned Bass so the low end softens and rests when you want the breathing to step back.
Variations:
- Disable the PWM LFO and the patch becomes a static square, the character is entirely the pulse-width motion.
- Add a slow filter LFO at 0.05 Hz, ±400 Hz, and the bass also slowly opens, moving toward a Floating Points "Silhouettes" register.
In the lineage of:
How to make a eurodance detuned-saw pluck with a snapping filter
Eurodance Saw. A bright detuned-saw pluck, one filter envelope snapping open and shut.
A bright detuned-saw pluck, the whole sound a single filter envelope snapping open and shut. It carried the Y2K dancefloor, all forward motion and chrome optimism, the riff that needs no melody beyond its own bounce.
What you'll need: You'll need a synth with a sawtooth oscillator, a wide unison, a sine sub, a lowpass filter with its own envelope, chorus, and a little drive. Most synths have all of these.
The recipe
- A wide stack of saws. A sawtooth on five-voice unison, detuned wide and the blend pushed up, so the stack is thick and chrome-bright before the filter ever touches it. The result is a wide shimmering swirl with no single pitch standing out; if it sounds like one clean saw, the detune or the voice blend is still too low.
- A low resonant lowpass. Close the lowpass down to around 600 Hz with moderate resonance. On its own the stack now sounds muffled; that closed state is the bottom of the envelope you are about to draw.
- The filter snaps open, the pluck. Drag an envelope to the cutoff with a large positive amount so the filter snaps open to around 4 kHz on the attack, then a fast decay back to closed. That snap of brightness is the pluck. You'll know the amount is right when each note opens with an audible chirp of brightness that drops away at once; too much and the snap whistles, too little and the note stays dull and never plucks.
- The amp dies with the filter. Match the amp envelope to the filter: fast attack, a decay of about 180 ms, no sustain. The note ends as the brightness does, which is what makes it a pluck and not a held lead.
- Chorus and a little drive. A wash of chorus for the lush Y2K width and a touch of filter drive for grit, the chorus spreading the image as its short modulated delays decorrelate the copies. Then play it as a steady, syncopated riff; the bounce is the hook. The stack should sit wide and glassy with a faint edge of grit on the attack; if it starts to sound harsh or smeared, pull the drive or the chorus back.
Add a quiet sine sub at the same octave so the pluck has a body under its brightness, then run the line in A3 to C4. The filter does all the moving; the notes barely change.
Variations:
- Lengthen the filter decay and it opens into a held trance lead; shorten it and it tightens into a stab.
- Drop the resonance and it smooths toward a plain saw pluck; raise it and the snap starts to whistle.
In the lineage of:
- Alice Deejay (Better Off Alone)
- Ian Van Dahl (Castles in the Sky)
How to make a deep-house chord pad
Deep House Pad. The chord wash that holds the floor together.
The pad that holds the room while the kick holds the floor. A wide, slow saw chord wash that breathes with the kick.
What you'll need: You'll need a synth with a unison control, enough polyphony to hold a chord, an amp envelope, and a compressor. All commonplace.
The recipe
- Five soft saws, gently detuned. OSC A on soft saw, unison 5 voices, detune ±7 cents. The pad is wide; the chord should feel like several voices holding the same line.
- A 600ms attack. AMP ATK 600ms, full sustain, REL 2 s. The pad blooms slowly enough that chord changes don't click but quickly enough that the kick stays in front.
- Slow filter LFO. LFO 1 on filter cutoff, sine at 0.08 Hz, depth ±400 Hz around the 2.8 kHz center. The pad slowly opens and closes, barely audible, but the patch feels alive over minutes.
- Sidechain to the kick. Compressor on the pad, sidechained to the 4/4 kick. Gentle pumping, the chord breathes with the floor. The tell is the chord swelling back between kicks without ever fully vanishing; if it gulps or stutters at the bottom of each duck, the sidechain is too deep.
- The sidechain IS the song. 4:1 compression keyed off the kick, 2–3 dB duck on every quarter note, high-pass at 80 Hz so the kick punches through, and gentle parallel reverb at 4 s decay on its own bus so the pad swells without fighting the kick. The kick lands clean and dry while the pad swells wet and wide behind it; if the room blurs into the kick, clear the reverb bus below 80 Hz.
Hold a chord. Change every eight bars. The kick does the rest. It holds the room behind the kick, so sidechain it gently and keep it wide and back so the floor stays in front. When you want the top to glow, float a String Section an octave above the chord at a whisper, voiced to the same changes, and the wash leans bright without crowding the floor.
Variations:
- Disable the sidechain and the patch becomes ambient, house without the pump is the chord without the room.
- Add a second pad an octave up at 30% mix and the patch leans bright, closer to Floating Points than to Larry Heard.
In the lineage of:
- Larry Heard (Can You Feel It)
- DJ Sprinkles (Grand Central Pt. 1)
- Floating Points (Silhouettes)
How to make a detuned disco chord stab
Detuned Disco Stab. Two slightly-detuned saws as a chord stab.
A bright chord flicked on the upbeat and gone before you can hold it. The two-saw stab, disco's gift to house.
What you'll need: You'll need a synth with a unison (voice-stacking) control, a lowpass filter, a short stab envelope, and reverb. All standard.
The recipe
- Two saws, 10 cents apart. OSC A and OSC B both on saw, detuned 10 cents. The slight beat between the voices is the analog character that makes the stab feel hand-played. Listen for a slow gentle shimmer between the voices, not a fast warble; if it beats quickly enough to sound out of tune, the detune is too wide.
- A pluck envelope. AMP ATK 2ms, DEC 180ms, SUS 0%, REL 80ms. The stab is gone almost before it lands. You'll know it is right when the note flicks and clears with no tail left behind; if any sustain hangs on, the decay is too long for a stab.
- Filter with resonance. Lowpass at 4.5 kHz, resonance 0.4. The resonance highlights the cutoff, giving the stab its signature bite. A sharp ringing edge rides the top of each stab; push the resonance higher and that edge tips into a whistle that overtakes the chord.
- Short reverb. Reverb decay 2 s, wet 30%. The stab needs space without losing the brevity that makes it a stab.
- Mix the stab for the dance floor. 4:1 compression with fast attack (the stab's transient should pop); a 2 dB boost at 3 kHz for presence; reverb pre-delayed by 50 ms on a parallel bus so the dry stab speaks before the room replies.
Place the stab on the and-of-2 and and-of-4. The bar swings. Bright and brief, it rides the upbeats, so keep it thin and panned with a short tail so it pushes the groove without crowding it. Underneath, a plain bassline landing square on the downbeats holds the low end while the stab answers it off the beat.
Variations:
- Raise the resonance to 0.7 and the patch becomes a bright filter-stab, closer to acid house than to deep house.
- Lengthen the decay to 400 ms and the stab becomes a chord chime; the off-beat lesson is in the brevity.
In the lineage of:
How to make a gliding late-set house lead
Late-Set Lead. Single-voice saw lead, soft attack, glide between notes.
A single saw voice gliding between notes, soft-edged and unhurried, the producer arriving at hour seven.
What you'll need: You'll need a synth with a mono (single-voice) legato mode, an amp envelope, and reverb. Nothing exotic.
The recipe
- A single saw, mono mode. OSC A on saw, one voice, mono mode. The lead is a single line; polyphony would make it a pad. The tell of true mono is that fast repeated notes never stack into a chord, only ever one voice sounding at a time.
- Portamento at 150ms. Portamento 150ms, legato. When notes overlap, the pitch glides from one to the next; when they don't, each note arrives clean.
- Soft attack. AMP ATK 60ms, not a pluck, not a pad. The lead settles in rather than striking. Stretching the amplitude ramp past the ear's click threshold smooths away the percussive transient an instant attack would leave.
- Moderate chorus and reverb. Chorus depth 30%, reverb decay 3.5 s, wet 45%. The lead sits comfortably in the room without insisting on the front. Raising the wet level relative to the dry signal lifts the reflected energy the ear uses to judge distance, seating the voice deeper in the space.
- Mix the lead as a single thread. Mono, no widening, gentle parallel chorus at 25% mix, a high-shelf of +1 dB at 4 kHz for presence over the floor, plate reverb at 3.5 s decay on the wet bus, dry path uncompressed. You'll know the balance is right when the lead reads clearly over a busy floor yet still feels like a single thread down the centre, not a widened wash.
Play a melody with overlapping notes, the portamento writes the line. Soft-edged and singing, it takes the front quietly, so keep the bed thin and let the glides carry the phrasing. A Mono Synth Lead makes the natural partner, the two trading phrases, one voice answering the other across the quiet floor.
Variations:
- Disable portamento and the patch becomes a generic saw lead, the glide is what makes it late-set.
- Drop the chorus depth to 10% and the patch becomes sharper, closer to a modular lead than an analog house lead.
In the lineage of:
How to make a minor-seventh Rhodes chord for deep house
Back Booth Rhodes. A minor-seventh chord from the corner nobody is dancing in.
A minor-seventh chord laid down in the corner booth, warm and a little worn, the comp underneath everything else in the room. An FM electric piano at a 1:1 ratio, a short bark of modulation on the attack landing into a soft, sustaining body.
What you'll need: You'll need a synth that can do FM, one sine modulating another, with an envelope on the FM amount, plus chorus. Most FM-capable synths cover all three.
The recipe
- An FM electric piano at a 1:1 ratio. OSC A on sine, OSC B modulating it at the same frequency, at a modest depth around 22 percent. The 1:1 ratio is the warm electric-piano register, rounder than the brighter bell you get from wider ratios. An integer ratio lays the FM sidebands on whole-number harmonics that fuse into one pitched tone, where wider ratios scatter them into metallic partials.
- The tine bark on the attack. A short burst of extra modulation at the very start, decaying over about 110ms. This is the tine bark, the knock of the hammer just before the tone settles into its body. The brief spike in modulation index throws up a spray of high sidebands that fall away as it drops back.
- A minor-seventh comp. Voice the chord as a minor seventh: root, flat third, fifth, flat seventh. Hold it. The chord is the comp the whole track leans on. You should hear the flat seventh leaning the chord just short of resolved, a held question rather than a settled answer.
- Bark, then body. Attack 6ms, a long decay down to a half-level sustain, release near a second. You hear the bark land and then the warm body hold under it.
- Chorus and a plate, set back. A little chorus, around 35 percent, for the worn stereo width, then a medium plate reverb. Present, but sitting at the back of the room rather than the front. The chorus widens by running short modulated delays that decorrelate the copies, so the two sides no longer line up.
Comp a slow minor-seventh pattern behind a track that is already running. Lay back; you got here hours ago. A comp sits in the middle behind everything, so keep it warm and low in level and let it fill space without stepping on the lead. The low end it leaves open wants an Organ Bass underneath to carry it.
Variations:
- Raise the FM depth and the tine sharpens toward a bell, a brighter room and an earlier hour.
- Lengthen the release until the chords overlap and the comp turns into a pad.
In the lineage of:
How to make a 90s house drawbar organ bass
Organ Bass. An additive drawbar organ with a percussive ping, the bass of 90s house.
The drawbar organ bass of 90s house, built from the ground up. A stack of sine harmonics is the organ tone, a sine an octave below is the weight, and a bright overtone with a click on top is the percussive attack that makes it read as a keyed instrument rather than a synth.
What you'll need: You'll need a synth that can stack sine harmonics (an additive or wavetable oscillator, or several sine voices), a sine sub an octave down, a lowpass to round the top, a fast bright envelope for the percussion ping, a noise burst for the key click, and chorus. Most modern synths cover all of these.
The recipe
- The drawbar frame. Build the tone additively: a sine fundamental with quieter sines at the octave, the octave-and-fifth, and two octaves up, plus a sine an octave below the root for the sixteen-foot weight. No detuning; a real organ does not chorus its own pipes.
- Round the very top. A lowpass around 1.5 kHz rounds the highest harmonic so the stack reads as an organ rather than a buzz. The tone lives in the drawbars; the filter only tidies the top.
- The percussive attack. An organ is keyed, not swelled. A fast attack, around 2 ms, decaying in about 100 ms to a held sustain, gives the struck front. On top of that, a short bright overtone, a high partial near 3.5 kHz that pings and dies in about a tenth of a second, is the signature percussion of the workstation organ. You'll know the ping is set right when you hear the attack as a separate bright tick over the note, not just a louder start.
- The key click. A tiny burst of filtered noise at the very start, twenty milliseconds and gone, is the key click that makes the note feel pressed rather than faded in. A touch of saturation behind it adds the keyed bite.
- Chorus, then bounce it. A moderate chorus, wet enough to widen but not so much it smears, for the 90s sheen, then play it short and staccato in the low octaves. The notes are clipped and repeated; the bounce is the bassline. Set the chorus so the organ widens and shimmers while a clear pitch still sits on every note.
Play it short and staccato from E1 to A1 and let the offbeats carry the groove. It is bright and present where a sub would disappear, the bassline doing the work a melody might. A Deep House Pad over the top gives it something to punch through; the pad holds the room while the bass does the walking.
Variations:
- Pull the higher drawbars and it darkens toward a rounder sub organ; add them back and brighten the percussion for the full keyed snap.
- Lengthen the decay and it sustains into a held organ; shorten it and gate it hard for a tighter, more percussive bounce.
In the lineage of:
- Robin S (Show Me Love)
- Crystal Waters (Gypsy Woman)
How to make an acid bass, the squelching 303 slide
Acid Bass. A resonant saw that squelches and slides, hypnotic by repetition.
One sawtooth, one resonant filter, and a hand on the cutoff. The whole sound is a single low line that opens and closes on itself, the squelch a small bass machine made famous when producers found it by misreading the manual. Hypnotic by repetition, not by melody.
What you'll need: You'll need a synth with a sawtooth oscillator, a lowpass filter with high resonance and its own envelope, portamento for the slides, and a little overdrive. Most synths have all of these.
The recipe
- One sawtooth, mono. A single sawtooth oscillator, no unison, no second voice. The 303 ran on one oscillator, a saw or a square, and the saw is the brighter, more common acid voice. The raw tone gives it away: a saw already buzzes bright and thin, with no hollow scoop the square would add.
- Resonance, turned up. A lowpass with the resonance pushed high, so the filter begins to ring, a whistling peak right at the cutoff. That ringing peak is the source of the squelch.
- An envelope on the cutoff, the squelch. Drag an envelope to the cutoff: a fast snap open, a short decay back down. The cutoff opening and closing once per note is the squelch, the eeyow that names the sound.
- A plucky gate. A fast attack and short decay on the amplitude, so each note is a stab rather than a held tone. The acid line is all short, clipped notes; the filter does the singing.
- The slide. Set the voice to mono with a short portamento, a glide of 40 to 80 ms. On a 303 the slide only fires between overlapping or accented notes, so the pitch sweeps from one to the next instead of re-striking. You'll know it is right when a slid note arrives by bending up or down into pitch, not by a clean new attack.
- Overdrive, then repeat. A little overdrive for grit, then loop the line and leave it. The hypnosis is in the repetition, not the notes; the filter moving under an unchanging pattern is the whole performance. The drive is right when it adds a dirty edge that thickens the squelch; too far and the resonant peak smears into a flat buzz and stops singing.
Run it as a looping sixteenth-note line and let a few notes slide into the next, the 303's signature slur. The pattern barely changes; the filter does all the moving, and the repetition is the point. From here, lay a Detuned Disco Stab over the loop on the offbeats, a little chord answering the squelch underneath.
Variations:
- Back the resonance off and shorten the envelope and it tightens into a plain plucked bass; push both and it screams, self-oscillating into a lead.
- Slow the line and stretch the decay for a dubbed-out, hypnotic acid that sways rather than drives.
In the lineage of:
- Phuture (Acid Tracks)
- Hardfloor (Acperience 1)
How to make a dream-trance mallet, a bright bell into a dreamy tail
Dream-Trance Mallet. A round mallet with a bright strike, ringing out into a dreamy tail.
A round core with a bright strike on top, ringing out into a dreamy tail. The body is a sine, the attack is FM, and the long reverb is the dream in dream trance.
What you'll need: You'll need a synth that can do FM, a fast amp envelope with no sustain, an envelope on the FM amount for the strike, a long reverb, and chorus. Most FM-capable synths cover all of these.
The recipe
- FM gives the strike. A sine carrier is the round body; a sine modulator at a non-integer ratio adds a burst of bright, slightly metallic overtones on the attack, the mallet hit. The tell is a glint of metal on the attack over a body that stays pure and round underneath.
- Struck, not held. Fast attack, fast decay, no sustain. A mallet has no body to hold it; the note is a strike and a ring-out, nothing in between.
- Tune the brightness. The FM amount sets how bright and metallic the strike is: more for a glassy bell, less for a soft mallet. Put it on its own fast envelope so the brightness dies before the note does, which is true bell behaviour. In real bells the high partials radiate and damp away faster than the fundamental, so the tone dulls as it rings.
- A round low body. Under the strike, a sine an octave down gives the big round body dream trance leans on, the weight the bright attack rides. Roll the very top off so the bell stays sweet rather than glassy.
- The dream: chorus and reverb. A lush chorus and a long reverb tail are what make it dream trance rather than a dry mallet. Let the ring-out blur into the room and the space between notes do half the work.
Play it high, C4 to E4, as a sparse melodic figure and let the reverb carry the space between notes. It is a lead that sounds like a memory of one. A Screensaver Pad underneath will hold the room while the mallet rings out over the top.
Variations:
- Drop the FM amount and lengthen the decay for a soft marimba; raise it and shorten the decay for a glassy bell.
- Add a sine sub an octave below and it gains the full dream-trance body; take it away and it floats.
In the lineage of:
- Robert Miles (Children)
- Chicane (Saltwater)
How to make a hollow square-wave pluck
Square Pluck. A woody square pluck, rounder than a saw, hollowed by its odd harmonics.
A woody, rounded pluck. A square wave's odd-only harmonics give it a hollow, clarinet-like tone for free, softer than a saw, and a filter envelope turns it into a pluck.
What you'll need: You'll need a synth with a square or pulse oscillator (PWM is a bonus), a sine sub, a lowpass filter with its own envelope, and chorus. Most synths have all of these.
The recipe
- A detuned square. A square oscillator on two or three voices, detuned a little. The square carries only odd harmonics, which is the hollow, woody character a saw cannot give. A slow LFO on the pulse width keeps it alive.
- A resonant lowpass, closed. Close the lowpass to around 500 Hz with moderate resonance. The square goes muffled and round, the bottom of the pluck envelope.
- The filter snaps the pluck. An envelope on the cutoff opens it to around 2.6 kHz on the attack and decays back, about 200 ms. That snap of brightness is the pluck.
- A plucky amp. Fast attack, a decay of about 200 ms, low sustain, so the note is a pluck with a little body, slightly longer than a mallet but still short.
- Sub for body, chorus for air. A quiet sine sub an octave down gives the hollow square some weight, and a touch of chorus and light drive round it off. Play it as a melodic pluck around A3. The sub and chorus should add weight and width while every note still reads as a clear single pitch, not a blurred wash.
Rounder and woodier than the eurodance saw, it sits well as a melodic pluck or a soft bass. Keep the resonance moderate so it stays warm rather than nasal. Set an Organ Bass beneath the pluck and the low end walks while the square carries the melody.
Variations:
- Shorten the decays for a tight, mallet-like blip; lengthen them for a held, woody lead.
- Modulate the pulse width faster and it thins toward a reedy, nasal tone; slow it and it stays round.
In the lineage of:
- Orbital (Belfast)
- Plaid (Eyen)
How to make a carillon, a tower of tuned bells
Carillon. A tower of tuned bells ringing out over a town that may not be there anymore.
A carillon is a tower of tuned bells, and a bell is the clearest case of an inharmonic tone: struck, it throws off partials that refuse to line up into a tidy harmonic stack, which is why it shimmers and beats instead of sitting still. Two sine operators at a 1.41 ratio build that metallic spread, struck once and left to ring.
What you'll need: You'll need a synth that can do FM with one sine modulating another at a non-integer ratio, an envelope on the FM amount, a filter envelope, and a long reverb. Most FM-capable synths cover all of these.
The recipe
- A two-operator FM pair at an inharmonic ratio. OSC A sine carrier, OSC B sine modulator at a 1.41 ratio. A whole-number ratio would give a harmonic, organ-like tone; the non-integer ratio scatters the sidebands inharmonically, which is what the ear reads as struck metal.
- Struck, then left to ring. AMP ATK 2ms, a long decay near five and a half seconds down to silence, no sustain. The envelope is the whole bell: an instant strike and a slow ring-out. Hold the note or let it go, the tail is already decided.
- The high partials die first. A filter envelope that eases the lowpass down over the first second and a half, so the bright clangor of the strike settles into a purer hum. A real bell sheds its top partials fastest; closing the filter slowly imitates that decay of brightness.
- Place it in the courtyard. A long, bright plate reverb near four and a half seconds, on its own bus so the strike stays defined while the tail spreads out into stone and air. Play single notes, slowly; a carillon is heard one bell at a time, from a distance.
Strike one note and wait. The bell does the rest, and the longer you leave it, the more it sounds like memory. Inharmonic and ringing, bells fight tuned chords, so give the carillon its own register high up and keep pads out of its way. A sparse Music Box Pluck in the same high register can answer the slow bells, struck between the rings.
Variations:
- Widen the ratio toward 2.76 and the bell turns to glass, brighter and thinner, a smaller bell heard up close.
- Shorten the decay to a single second and the carillon becomes a music box, struck and gone before it can ring.
In the lineage of:
How to make a gamelan, shimmering struck bronze
Gamelan. Tuned bronze, struck and shimmering, gone before it settles.
A gamelan is tuned bronze, and bronze rings inharmonically: struck, it throws off a bright spray of partials that fade fast and unevenly. Two sine operators at a 1.62 ratio build that shimmer, a lighter, shorter cousin of the carillon, struck and already leaving.
What you'll need: You'll need a synth that can do FM with one sine modulating another at a non-integer ratio, a short percussive amp envelope, a lowpass filter, and reverb. Any FM patch handles this.
The recipe
- An inharmonic FM pair, bronze ratio. OSC A sine carrier, OSC B sine modulator at a 1.62 ratio, a modest modulation index. That non-integer ratio scatters the partials inharmonically, which is what the ear hears as struck metal. Aim for bronze, bright and slightly clashing; too much index and it rings on into a tuned bell.
- Struck and gone. AMP ATK 3ms, decay near two seconds to silence, no sustain. Shorter than a bell, longer than a marimba; the bronze rings briefly and then is simply not there.
- Keep the bronze warm, not brittle. A gentle lowpass near 9 kHz. The strike keeps its bright edge, but the most brittle top is rolled off so the tone reads as metal in a room rather than a digital ping.
- Set it ringing in a hall. A light chorus for width, a long reverb of around a four-second tail for the hall, and alternate notes panned about thirty percent left and right the way a player moves across the bars. Strike a slow pentatonic figure and let the tones overlap. You'll know the reverb is right when the strikes blur into one another without turning to mush; you should still hear each note arrive.
Play a slow five-note figure and let the strikes ring into each other; the bronze writes the harmony for you. Inharmonic bronze clashes with tuned pads, so give it its own space high in the mix and keep harmony instruments clear of it. Under it, a Mall Marimba an octave down gives the soft attacks and lets the bronze answer from above.
Variations:
- Drop the ratio to 1.41 and the bronze grows into a bell, longer and heavier, the carillon in the next tower over.
- Stretch the decay past three seconds and a struck gong appears, the same metal heard from much further back.
In the lineage of:
- Visible Cloaks (Reassemblage)
- Midori Takada (Through the Looking Glass)
- Haruomi Hosono (Watering a Flower)
How to make a parlor harmonium with pump-bellows tremolo
Harmonium. Reed-organ chords with a slow pump-bellows tremolo, a front-room hymn.
A parlor harmonium makes its sound by pulling air across metal reeds, so the tone is buzzy and a little unsteady, and the foot-pumped bellows give it a slow breathing tremolo. Three square-wave voices a few cents apart build the reed buzz; a gentle wobble in the amplitude is the bellows.
What you'll need: You'll need a synth with a square or pulse oscillator, a unison control, a lowpass filter with a small resonant peak, an LFO routed to the amplitude for the tremolo, and reverb. Most synths have all of these.
The recipe
- Three reed voices, slightly out of tune. OSC on square, unison 3 voices, detune spread about 18 cents. Square gives the buzzy, odd-harmonic body of a reed; the slight detune is the hand-pumped reeds never quite landing in tune with each other. The chord should shimmer and beat slowly, unsettled, without tipping into sounding broken.
- The body of the box. A lowpass near 2.6 kHz to tame the square buzz, plus a small peak around 1.3 kHz for the wooden cabinet the reeds sit in. The peak is the difference between a raw square and an instrument with a body.
- Soft in, soft out. AMP ATK about 140ms, REL about 900ms. The bellows take a moment to build pressure, so the chord swells in rather than snapping on, and falls away just as gently when you lift your hands.
- The bellows breathing. An LFO on the amplitude near five and a half cycles a second, depth shallow, moving maybe a sixth of the level. This is the give and take of the foot pump, a tremolo no one plays, the instrument breathing on its own under the chord. You'll know the depth is right when the chord pulses without ever stuttering to silence; it should wave, not gate. A little chorus and a small room finish it.
Hold a slow chord and let the tremolo breathe under it; the harmonium was built for hymns played alone in a front room. A reedy, unsteady tone owns the midrange, so let it sit there mostly alone, kept a touch back so the buzz reads as character rather than clutter. A Slow-Bloom Pad beneath it softens the edges while the harmonium stays forward as the voice.
Variations:
- Speed the tremolo past eight cycles a second and the parlor organ turns seasick, closer to a carnival than a chapel.
- Switch the reeds from square to a softer triangle and the buzz drops away, leaving something nearer a flute stop.
In the lineage of:
- Beach House (Devotion)
- Mazzy Star (Among My Swan)
- Cat Power (Moon Pix)
How to make a breathy new-age pan flute
Pan Flute. A breathy flute, more air than tone, lifted off a new-age relaxation tape.
A flute is nearly a pure tone with a lot of breath around it: a triangle wave carries the hollow, woody body, and a layer of filtered noise is the air pushed across the mouthpiece. A single voice, a vibrato that arrives only after the note has settled, and a breath that never fully leaves.
What you'll need: You'll need a synth with a triangle oscillator, a noise layer for the breath, a lowpass filter with a small resonant peak, an LFO for late-onset vibrato, and reverb. All common building blocks.
The recipe
- A single hollow voice. One triangle oscillator, no unison. A flute is a single player, and the triangle gives the hollow, harmonically sparse body a flute has, far closer than a bright saw or square. The tone should sit soft and woody with no buzz at its edge, the saw's reedy fizz gone.
- The breath. A noise layer, lowpassed around 3.8 kHz, low in the mix, riding under the tone. This is the air across the mouthpiece. Cut it and the flute goes instantly synthetic; a small parameter carrying most of the realism, because the ear listens for the breath every real wind instrument carries.
- A small tube resonance. A gentle peak around 850 Hz for the body of the pipe. It is a subtle bump, but it places the breath and the tone in the same hollow space rather than layering noise over an oscillator. You'll know it landed when the breath and the tone fuse into one hollow voice; nudge the gain past a few dB and the body starts to honk instead of sit under the air.
- Late vibrato, the player leaning in. Vibrato near 4.6 Hz at about 18 cents, onset near 480ms. The note arrives flat and steady, then the player leans in and the vibrato blooms. Vibrato from the very first instant sounds mechanical; the delay is what sounds human.
- Soft edges, a small room. A soft attack near 100ms and a release near 700ms so notes enter and leave on breath rather than a hard edge, then a small, short reverb. Play slow, stepwise melodies; the flute rewards space more than speed. Each note should fade up out of the breath rather than starting on a click.
Play a slow, stepwise line and let each note breathe in. More air than tone is the whole point. Breathy and soft, it sits forward but quiet, so keep the arrangement thin beneath it and let the air read as intimacy. A String Section underneath gives the thin pad bed the flute breathes over.
Variations:
- Push the breath layer up and the flute becomes a shakuhachi, more air than pitch, almost spoken.
- Remove the vibrato and the tone goes still and synthetic, a recorder rather than a flute.
In the lineage of:
How to make a wavering glass theremin, the sci-fi sine
Glass Theremin. A single sine that can't stop wavering, the ghost in every old science-fiction reel.
A theremin is the purest tone in music: a single sine with no harmonics, no edge, nothing but pitch and the hand that shapes it. Everything expressive lives in the vibrato and the slow swell in and out. Take almost everything away, and what is left has to carry the whole performance.
What you'll need: You'll need a synth with a sine oscillator, an amp envelope with a slow attack and release, an LFO routed to pitch for a wide late-onset vibrato, and a large reverb. All standard.
The recipe
- One pure sine. One sine oscillator, a single voice, nothing stacked. A sine is the only waveform with just one harmonic, so it has no buzz or edge at all. Next to a saw it sounds hollow and ghostly, which is exactly the theremin voice.
- Bow it in. A slow attack near 320ms so the note swells up from silence with no hard onset, and a long release so it sinks back the same way. A theremin has no key or pick; every note is bowed in and out by hand. The envelope is doing the bowing here. You'll know the swell is right when you cannot point to the instant the note began, only the moment it was suddenly there.
- The waver, arriving late. Vibrato near 5.6 Hz at about 55 cents, arriving near 650ms after the note starts. The waver is wide and a little fast, and it blooms in late rather than being present from the first instant. The hand of a theremin player never stops moving, and that restless vibrato is most of what makes the tone sound alive rather than like a test tone.
- A glass of air around it. A long, soft reverb to set the sine in a large empty space. With so little in the patch, the room becomes part of the instrument; the tail is where the eeriness lives. Play slow, sliding lines and let each note swell, waver, and fade. Deep enough, and the tail keeps wavering for a breath after your hand has left the note.
Play it slow and let your phrasing do everything. It is one sine wave; the performance is the patch. A bare sine reads as fragile, so keep it forward and exposed with almost nothing beneath it, placed in a room by reverb. Let a faint Ghost Melody drift in far beneath it, answering the theremin's slides from another room.
Variations:
- Narrow the vibrato and slow it down, and the theremin becomes a glass harmonica, still and bell-like.
- Open the attack even further and drop the pitch an octave, and it becomes the ondes Martenot under an old film.
In the lineage of:
- Clara Rockmore (Theremin Recital)
- Delia Derbyshire (BBC Radiophonic)
- Bebe and Louis Barron (Forbidden Planet)
How to make a cathedral pipe organ in open fifths
Cathedral Organ. Three pipe ranks in open fifths, a chord that fills a stone room with dread.
A pipe organ does not play chords so much as fill a room. Each note is really a stack of pipes: the note itself, the octave above, and the twelfth above that, all sounding together as one voice. There is no third in the stack, so the sound is open and vast rather than happy or sad, and the cathedral around it is as much the instrument as the pipes.
What you'll need: You'll need a synth that can stack oscillators at the octave and twelfth, a triangle wave for the pipes, a unison control, an LFO routed to the amplitude for the tremulant, and a large reverb. Most synths have all of these.
The recipe
- The principal rank. One triangle rank at the root. A flue pipe is close to a triangle wave, hollow and with few harmonics, nothing like a buzzy saw. Two voices a few cents apart give the gentle beating of two pipes that are never perfectly in tune. The held note should carry a slow, breathing waver, not a fast warble or a dead-still tone.
- Stack the octave and the twelfth. Add a rank an octave up and a rank a twelfth up, the octave and the fifth above it. This is how a real organ builds a stop: parallel pipes tuned to the harmonics of the root. There is deliberately no third, so the stack stays open and cavernous instead of sounding like a major chord.
- The room is the instrument. A long reverb, several seconds of tail. A cathedral organ is unimaginable without the cathedral; the stone room blurs the ranks together and turns separate pipes into a single wall of air. Push the reverb until the room answers back as loudly as the pipes.
- A slow tremulant. A slow tremulant, a gentle waver in the volume near 4 Hz, the way a real tremulant trembles the wind supply. Keep it subtle. Play long, held chords low on the keyboard and let the room do the rest; this is a sound for dread and grandeur, not for fast playing. You'll know it is subtle enough when the chord seems to breathe rather than pulse; if you can tap along to the waver, pull the depth back down.
Hold the chord and let the room bloom. The pipes state it once; the cathedral sustains it. An organ fills the whole spectrum, so let it be bed and harmony both, and pull everything else thin while it holds. A Slow-Bloom Pad can swell up only in the silences between the chords, answering the organ without crowding it.
Variations:
- Drop the whole stack an octave and the organ becomes a 32-foot pedal rank, more felt than heard.
- Remove the twelfth and keep only octaves, and the open fifth gives way to a purer, cleaner principal.
In the lineage of:
- Kali Malone (The Sacrificial Code)
- Ellen Arkbro (For Organ and Brass)
- Charlemagne Palestine (Schlingen-Blängen)
How to make a smooth-jazz sax, the weather-channel sound
Smooth Sax. The smooth-jazz sax of a weather forecast, breathy and a little too relaxed.
The smooth-jazz saxophone is a reed sound that has been polished until all its edges are gone. It starts as a buzzy saw, the rich, all-harmonics tone of a vibrating reed, then a warm filter rounds off the top, a strong mid formant adds the nasal honk that says saxophone, and a breathy attack and a late vibrato make it sound played rather than generated. It should feel a little too relaxed.
What you'll need: You'll need a synth with a saw oscillator, a lowpass filter, a resonant peak (a band of EQ) for the formant, a noise layer for breath, an LFO for late vibrato, and reverb. Nothing unusual in that list.
The recipe
- A reed, not a pipe. A saw oscillator, one voice. A reed is rich and buzzy, full of harmonics, the opposite of a hollow flute. The saw gives the sax its body and brass; start here and everything after is about taming it. The bare tone should sound too bright and buzzy to be a sax yet, all reed and no polish.
- Round off the buzz. A lowpass around 3.2 kHz to roll off the harsh top of the saw. Smooth jazz is warm, never bright or buzzy; the filter is what turns a raw saw into something that sounds like it was recorded through a soft-focus lens in 1994.
- The reedy honk. A strong peak around 1.4 kHz. This is the formant, the fixed resonance of the saxophone body, and it is the single most important move: it is the difference between a filtered saw and a sound the ear hears as a reed. Sweep it a little and you can hear the vowel of the instrument change.
- A breath and a late vibrato. A light breath layer for the air across the reed, and a gentle, shallow vibrato near 4 Hz that arrives well after the note has spoken. Jazz vibrato is slower and warmer than a theremin waver, and a player adds it once the note is already singing, not from the first instant. Together they sell the performance. You'll know the vibrato is late enough when the note speaks flat and steady first, then warms into a slow waver a beat later.
- Subtone and a little room. A soft attack, a touch of saturation for reed warmth, and a small, plush reverb. Play slow and a little behind the beat, with long held notes; this is music for a five-day forecast, unhurried to the point of sleepy. The tell on the saturation is the reed turning warm and a little thick, never crackling or gritty at the edge.
Hold the long notes and let the vibrato arrive on its own. The sound is doing the relaxing for you. A polished reed sits in the upper mids, so let it lead over a sparse bed and keep other midrange out from under it. Underneath it wants a Cheap Rhodes comping soft chords, the smooth-jazz bed the sax was made to lead over.
Variations:
- Raise the breath layer and soften the attack further for a subtoned ballad sax, more air than reed.
- Open the filter and push the formant higher for a brighter, more honking alto that cuts through a mix.
In the lineage of:
- The Weather Channel (Local on the 8s)
- Sade (Smooth Operator)
- George Benson (Breezin)
How to make a CD-skip stutter effect
CD Skip. The disc hanging on the one song you knew by heart.
The disc catches, and a fragment of the song loops, the same instant repeating, digital and pitiless, until someone bumps the tray. A harsh FM tone chopped into a stutter, the sound of media failing in real time.
What you'll need: You'll need a synth that can do FM, a square LFO gating the amplitude (or a gate/trance-gate effect), a little saturation, and an open filter. Most synths have all of these.
The recipe
- A hard digital tone. FM with the modulator at a non-integer ratio and the index pushed up, so the overtones land inharmonic and metallic, the cold brightness of a digital read rather than a warm instrument.
- Chop it into a stutter. A fast amplitude gate, twelve to sixteen times a second with a short open time, chops the held tone into a rhythmic stutter: a square LFO on the amp, or a trance-gate, does it. That gate repeating one slice is the skip. You'll know the rate is right when it reads as a disc stuck on a loop, mechanical and even, not as a tremolo or a beat you would tap to.
- Keep it bright and dry. No reverb, almost no decay. A skip is present and merciless, right at the surface, not in a room. Leave the top end open so the digital edge stays sharp.
- A digital bite. A little saturation hardens the gate edges into clicks, the laser stumbling on a scratch. Subtle; the rhythm should still read as a loop, not a buzz.
- Lurch the loop. Every few bars, nudge the gate rate or jump the pitch up a step and back, so the skip lurches and re-catches instead of sitting in a clean loop. Small moves, one parameter at a time, until the glitch sounds like the disc fighting to recover rather than a part you programmed.
Use it as rhythm, not melody: a stuttering fragment under a beat, or the moment a track tears itself apart. A little goes a long way; let it be the thing that breaks. Smear an AM Radio Wash underneath and the skip stops being an effect and starts sounding like a real machine coming apart.
Variations:
- Slow the gate to a few hertz and it becomes a glitchy gate rather than a skip; speed it past 30 Hz and the stutter fuses into a buzzing tone.
- Drop the FM index and it softens toward a clean pulsed sine, more test-tone than failure.
In the lineage of:
- Oval (Systemisch)
- Ryoji Ikeda (Dataplex)
How to make a worn dictaphone voice, band-limited and thin
Dictaphone Voice. A voice on a worn tape, band-limited and barely there.
A voice gone to tape, then to a cheaper tape, then to a machine with a worn head. Band-limited to a phone-line slot, two formants giving it a vowel, wavering and degraded until the words are gone but the shape of speaking remains.
What you'll need: You'll need a synth with a saw or pulse oscillator, a narrow bandpass filter, a second resonant peak (a band of EQ) for a formant, a slow vibrato, and saturation. Most synths cover all of these.
The recipe
- A thin, reedy tone. A single sawtooth, rich in harmonics, is the raw material; a voice is just harmonics shaped by a throat. Keep it mono and plain for now: bright and buzzy with nothing vocal in it yet, just harmonics waiting to be shaped.
- Squeeze it to a phone line. A narrow band around the midrange, the way a telephone or a dictaphone throws away everything but the part that carries speech. Suddenly it sounds transmitted, not played, because the phone network only ever passed that narrow midrange slot, so the ear reads the absent lows and highs as the mark of a transmission.
- Two formants make it a voice. Add two resonant peaks, around 1.1 and 2.4 kHz, the formants that spell a vowel. They are the difference between a buzz and something almost speaking; the ear reads a mouth into them.
- Wow and degradation. A slow vibrato is the unsteady capstan of a worn machine, and a little saturation is the tape and the cheap preamp. Together they age the tone past clean into remembered.
- Let it waver and fade. A soft attack and a quick fade is the recording cutting in and out. Play short phrases, not held notes, and let each one sink back into the hiss it came from. You'll know it is right when each phrase seems to surface for a moment and then drop away, as if the worn machine keeps losing the thread.
It is not a sample and it will not say words; it is the ghost of a voice, the formants and the wow without the language. Sit it low under tape hiss and it reads as something half-remembered being said in another room. A Vocal Tape Drone hung beneath it gives the half-remembered voice its haunted bed to surface out of.
Variations:
- Widen the band and drop the saturation and it cleans up toward a nasal synth lead; narrow it further for a pure telephone tone.
- Raise the vibrato depth and it slips from unsteady-tape toward seasick; lower it and it steadies into a flat, dead recording.
In the lineage of:
- The Caretaker (Everywhere at the End of Time)
- Leyland Kirby (Eager to Tear Apart the Stars)
How to make a distant-highway drone, the far-off traffic wash
Distant Highway. The far-freeway wash that breathes as the cars pass.
The sound you only hear when everything else stops: a wall of far-off traffic, low and steady, swelling and ebbing as unseen cars pass. Filtered noise over a low rumble, breathing on its own.
What you'll need: You'll need a synth with a noise oscillator through a lowpass filter, a low sine for the rumble, a very slow LFO on the amplitude, and reverb. Most synths have all of these.
The recipe
- A low rumble. A single low sine near 49 Hz, a low G, is the body of the road, the part you feel before you hear. Roll a lowpass down to around 200 Hz so nothing but weight remains. You'll know it is sitting right when you can feel the rumble in your chest but can't quite hum its pitch; the moment it carries a clear note, it has come too close.
- The wash. A bed of white noise is the traffic itself, a thousand engines smeared into one tone. Run a noise oscillator through a lowpass and pull the cutoff down to around 1.4 kHz, so the hiss loses its edge and turns into air. Set its level so it sits just over the rumble, present but never hissy. Done right, it stops reading as noise and starts reading as distance, a wash you forget is a sound.
- Push it back. The darker the lowpass, the farther away the highway: distance is just the high frequencies the air took away. Around 1.4 kHz the road sits at the horizon, 3 kHz pulls it up close, under 800 Hz it slips to the edge of hearing. Move the one cutoff until the distance matches the one in your head, then leave it; that single control is the whole sense of space.
- Let it breathe. A very slow LFO on the volume, one cycle every eight seconds or so, swells and ebbs the wash like waves of passing cars. Keep the depth shallow, moving maybe a third of the level, not gating it on and off; the breath should be felt more than heard. The rate is right when you cannot predict the next swell: faster and it turns into a pulse, slower and it flattens back into stillness.
- Distance with reverb. A long, dark reverb sets the whole thing back behind glass: a decay of around four seconds, with the top of the tail rolled off so it never sparkles. Not a room you are standing in, a horizon you are looking at. The tell of far enough is a reverb with no front edge you can point to, only a depth the dry sound disappears into.
Lay it under everything at low volume and it becomes the room tone of a track, the suburban night the other sounds happen inside. Keep it dark and let it breathe. Slide the Underpass Sub into the low end beneath it, or a Streetlight Pad on top, and the two distances stack into a whole street.
Variations:
- Open the lowpass and raise the level and the highway moves closer, an overpass rather than a horizon.
- Speed the breathing LFO and the swells become individual cars; slow it to stillness and it flattens into pure distance.
In the lineage of:
How to make a night-hiss ambience, the suburban hush after midnight
Night Hiss. The broadband hush of a suburb after midnight, fluttering faintly.
The hush a suburb makes after midnight: a fine, broadband hiss with a faint flutter in it, the sound of air and insects and distant electricity. Almost nothing, but never quite silence.
What you'll need: You'll need a synth with a noise oscillator and a lowpass filter, a faint low sine, and a fast LFO on the amplitude for the flutter. Most synths have all of these.
The recipe
- A bed of noise. White noise through a lowpass left fairly open, up around 8 kHz, is the night air itself, broadband and soft. Roll only the very top off so it reads as a hush rather than a hard static, and keep its level low. It is sitting right once you stop hearing it as noise and start hearing it as the quiet of a room.
- A low hum underneath. A faint low sine, down near the floor of hearing, is the distant electricity, the felt weight under the hiss. Barely there, more pressure than pitch.
- The flutter. A faster LFO on the volume, a few cycles a second, depth shallow, around fifteen percent, adds the shimmer of insects and moving air, a flutter rather than a swell. Keep it a little irregular so it reads as a night, not a tremolo. Shallow enough means you feel the air move but the level never visibly pumps.
- Keep it shapeless. Keep it shapeless: no resonance, no pitch you can name, no rhythm you could count. The moment a note or a pulse emerges it stops being ambience and becomes an instrument. You'll know you have gone too far the second you can hum it or tap to it; pull that motion back out.
- A little space. A short, soft reverb gives the hush somewhere to sit without turning it into a room. Just enough that it feels outdoors.
It belongs under everything, the quietest layer, the one you only notice when it stops. Mix it almost too low and let it be the silence the track sits in. A Distant Highway laid under it for a far-off low end turns the two together into a whole suburb of air.
Variations:
- Brighten the noise and quicken the flutter and it leans toward an insect field; darken and slow it for wind and distance.
- Drop the hum and it floats free; raise it and the night gains a faint electrical weight.
In the lineage of:
How to make a granulated dream texture you can freeze
Dream Texture. A held chord granulated into a slow, shimmering cloud you can freeze.
A held chord fed through a granular cloud until it stops being a chord and becomes weather: shimmering, slow, never quite repeating. Freeze it on a single moment and it hangs there forever. The dream lives in the cloud, the space, and the drift, not in the notes.
What you'll need: This build asks more of your synth than most: a granular oscillator you can freeze, position and pitch randomization, a slow filter LFO, and a long reverb. Not every synth has a granular mode; the per-step audio shows where it is headed.
The recipe
- The seed. Open the granular oscillator and load a sustained source, a held synth chord, a sung 'ah,' a piano chord. The source decides everything downstream; pick one with harmony worth dissolving. You chose well if the held source already sounds lush enough that you would happily let it ring on its own.
- A smooth cloud. GRAIN DENSITY 25 to 40 Hz, LENGTH 150 to 250 ms, window Hann or Gaussian, never rectangular or the grain edges click. Dense enough to read as one sustained tone, but mind the voice budget so the grains do not audibly steal from each other.
- Air and shimmer. PITCH randomization 5 to 15 percent and PAN randomization for width, with a little position jitter; scattering grains to random pan positions decorrelates the channels into what the ear reads as stereo space. Keep the pitch jitter decoupled from grain length so adding shimmer does not change density. The shimmer is right when the cloud widens and glitters at the edges yet still holds its pitch; too far and it sounds detuned and seasick.
- The space. A long convolution reverb or a lush hall, the tail measured in several seconds. This is the ambient hallmark: the cloud is the source, the reverb is the dream it floats in. A slow LFO on the filter cutoff underneath keeps it breathing. The space is deep enough once the dry cloud all but disappears into the tail and you can no longer tell where the grains end and the room begins.
- The freeze. Set SCAN to 0, or switch to manual mode, to stop the playhead and freeze the grains on a single instant. The cloud suspends, one shimmering moment held past time. Map it to a macro or the mod wheel; this is the signature gesture. You'll know the freeze took when the texture stops evolving and hangs on one unchanging shimmer that never moves forward.
- Tape, drift, and macros. A touch of tape saturation and a slow LFO on pitch add the wow and warm decay; an optional tiny frequency shift gives the wrong-but-musical drift, offsetting every partial by the same hertz rather than the same ratio so the harmonics slip out of whole-number lock. Then map macros, bloom, freeze, shimmer, age, and keytrack it so it plays across the keyboard. The tell is the cloud slipping out of true and never quite settling, wrong in a way that still sounds musical.
Map a bloom, a freeze, and an age control to macros and it stops being a patch and becomes an instrument: a five-minute drone that never reveals a loop. Let the reverb run long and resist adding notes; the texture is the piece. If one melody must surface, let a sparse Ghost Melody rise faintly from inside the cloud and sink back, so the texture stays the piece.
Variations:
- Smooth drone: dense grains, no freeze, a slow scan, an endless evolving wash.
- Glitch fragments: longer grains and heavy position and pitch randomization, the source breaking into shards.
- Frozen bloom: freeze the scan, then open the filter and reverb on a macro, one held moment that swells.
- Vocal ghost: granulate a sung 'ah' instead of a chord and the cloud takes on a wordless, human shimmer.
In the lineage of:
How to make a shellac 78 sound, buried in crackle and wow
Shellac. A 78 buried in crackle, band-limited and wowing on a tired platter.
An old pressing of a tune nobody kept. Band-limited to the middle of the spectrum, wowing on a tired platter, the surface noise as loud as the music. The decay is the recording, not the song.
The recipe
- The source. Start from a held chord, something with a little harmony. The melody barely matters here; what you are making is the recording of it, not the performance.
- Band-limit it hard. A bandpass parked in the midrange, roughly 300 Hz to 3 kHz, the lows and the air both gone, about telephone width. This one move is most of the age: a 78 had almost no bottom and almost no top. Narrow enough, and the music sounds like it is coming through a wall, present but with no weight and no sparkle.
- Wow. A slow, shallow pitch wobble, an LFO on pitch around half a cycle a second, only a few cents deep, the platter slightly off center. Right is when the pitch feels unsteady but never sounds out of tune; a little reads as a tired turntable, a lot reads as seasickness.
- Crackle and hiss. Two layers: a steady high hiss, the needle riding the groove, and random pops over the top, denser the more worn the disc. Set the pops as loud as the music itself; that imbalance is the point. You'll know it is right when the noise stops feeling like a layer on top and starts feeling like the recording you are listening through.
- Resist cleaning it up. No wide reverb, no stereo, no polish. The mono, the narrow band, and the noise floor are the sound. A short, faint room at most.
Hold one sustained chord and it should read as a memory of music rather than music: present, worn, a little sad. The less you add, the older it sounds. Loop a two-bar fragment and let it repeat past the point of comfort, the way The Caretaker does; the repetition is what turns a worn chord into hauntology.
Variations:
- Pre-war ballroom: narrow the band further and slow the wow.
- Field cylinder: more hiss, fewer pops, even less top end.
- Worn 45: widen the band a touch and speed the wow for a later, brighter pressing.
In the lineage of:
How to make a dying-Walkman effect, pitch and flutter sagging
Dead Battery. The Walkman giving up: pitch and tempo sagging, the flutter widening, the highs going out.
A tape playing as the battery dies. Pitch and tempo sag together, because it is the transport slowing, not a pitch knob. The flutter widens into a warble, the top end goes out, and then it is gone.
The recipe
- The source. Start from a warm held chord that already sounds like a dub of a dub, soft at the top before anything has happened to it.
- Sag pitch and tempo together. The tell of a dying transport: pitch and speed fall as one. Glide the whole thing down together over about eight seconds. You'll know it is right when it sounds like the machine slowing, not a pitch knob turning; if only the pitch sagged, the steady tempo would give the lie away.
- Widen the flutter. A fast pitch wobble, a few cycles a second, whose depth grows as the tape fails, the worn capstan losing its grip. Start it near nothing and let it widen to an audible warble by the end. It is working when the warble arrives on its own as the sound sags, rather than wobbling evenly from the start.
- Close the highs. A lowpass sweeping shut as the power fades, opening bright and closing down toward a few hundred hertz by the end. The top end is the first thing to go. The tell on the sweep is the sound seeming to move further away as it dies, not just get quieter.
- Let it die. A final fade as the motor stops, with a steady tape hiss underneath that outlives the music by a moment.
One press is one death. The drain time sets how long it takes; the flutter sets how sick the motor gets. Do not loop it, the ending is the piece. Record one full death as audio and drop that clip into a track as a one-shot, so the collapse lands exactly once where you want it.
Variations:
- Quick death: a short drain, the sound collapsing in a couple of seconds.
- Slow fade: a long drain for a tape that takes its time forgetting.
- Merely tired: low flutter and a gentle sag for a player worn, not dead.
In the lineage of: