REFERENCE · a series compiled by structure-void · card 01 · rev. 02
AUDIO EFFECTS / DECONSTRUCTED
A working reference for the twenty processors that shape every record you've ever heard. Each block lists what the effect actually does to the signal, the parameters that matter, a typical operating range, and the gotcha that ruins mixes. No best-fors. No emoji. No magic.
↓Get this card as a PDF — print-ready A3, workshop or classroom format.
FRQ — frequency
DYN — dynamics
HAR — harmonic
TIM — time-based
MOD — modulation
PIT — pitch / time
UTL — utility / image
/ 01—04
FREQUENCY DOMAIN
selectively amplify or attenuate spectral regions
/ 01FRQ
EQUALIZERlinear / minimum-phase / dynamic
Boost or cut narrow or broad frequency bands using shelves, bells and high/low-pass filters. The first decision in any chain — sculpt before you compress.
A compressor whose detector listens only to the sibilance band. When "ssss" or "ts" peaks, gain reduction triggers — but only there, and only then. Surgical, not destructive.
FREQ
4 – 10 kHz
RANGE
−2 to −10 dB
MODE
wideband · split-band
DETECT
side-chain HPF
ABLETON
Multiband Dynamics · Glue Comp + sidechain HPF
MAX
compressor~ + biquad~ (BP en sidechain)
gotcha · over-compressed sibilance becomes a voice without presence.
/ 03FRQ · UTL
HP / LP FILTERroll-off boundaries
Strictly speaking part of EQ, but worth its own card. HP filters remove rumble, sub-content and proximity build-up; LP tames brittle highs. The two filters that fix more mixes than any plugin.
SLOPE
6 / 12 / 24 / 48 dB·oct
HP
30 – 120 Hz typical
LP
10 – 18 kHz typical
RES
0 – 1.5 (Q at corner)
ABLETON
Auto Filter · EQ Eight (modes HP/LP)
MAX
lores~ · reson~ · onepole~ · svf~
gotcha · near the cutoff, steep slopes produce audible resonance.
/ 04FRQ · UTL
SPECTRUManalysis, not effect
Not an effect — a mirror. FFT-based displays expose what the ear half-hears: room modes, masking, spectral imbalance. Every serious chain begins and ends with one of these open.
FFT
1024 – 16384 samples
WIN
Hann · Blackman · flat-top
WEIGHT
flat · A · K
SLOPE
0 / 3 / 4.5 dB·oct (pink)
ABLETON
Spectrum · Tuner
MAX
fft~ · pfft~ · spectroscope~
gotcha · the trace is a snapshot in time; the ear integrates motion.
/ 05—09
DYNAMICS
amplitude as a function of itself
/ 05DYN
COMPRESSORdownward · feed-forward
Attenuates audio above a threshold by a chosen ratio, restoring perceived loudness with make-up gain. The transfer function is a kink, not a curve — though the knee softens the bend.
THR
−40 to 0 dBFS
RATIO
1.5:1 – 20:1 (∞ = limiter)
ATK
0.1 – 100 ms
REL
5 ms – 5 s
KNEE
0 – 18 dB
ABLETON
Compressor · Glue Compressor · Multiband Dynamics
MAX
omx.comp~ · slide~ · peakamp~
gotcha · a short attack time intercepts the transient before it exists.
/ 06DYN
LIMITERcompressor at ∞:1 — true-peak
A compressor with infinite ratio and very fast attack. The signal does not exceed the ceiling, period. Modern limiters predict inter-sample peaks and look ahead by milliseconds to avoid distortion.
CEIL
−1.0 to −0.1 dBTP
GAIN
0 – +18 dB input
LOOK
1 – 10 ms
REL
auto · 5 ms – 1 s
ABLETON
Limiter
MAX
omx.peaklim~ · clip~ + lookahead delay
gotcha · beyond a few dB of reduction, the limiter no longer contains — it distorts.
/ 07DYN
NOISE GATEdownward expander, hard
Below the threshold, the gate closes — silence. Above, it opens. Brutal but invaluable for kicks, snares and live mics that bleed everything else into the take.
THR
−60 to 0 dBFS
HOLD
0 – 500 ms
ATK / REL
0.01 – 200 / 5 – 2000 ms
RANGE
−∞ to 0 dB (depth)
ABLETON
Gate
MAX
gate~ · thresh~ · peakamp~ + gate~
gotcha · on a signal floating around the threshold, the gate oscillates between open and closed.
/ 08DYN
EXPANDERdownward — gate's polite cousin
Like a compressor, but applied below threshold and inverted: the quiet bits get quieter. Restores natural dynamics in over-compressed material, or thins out room tone without slamming the door.
THR
−60 to −20 dBFS
RATIO
1:1.5 – 1:8
ATK / REL
0.1 – 50 / 50 – 1000 ms
RANGE
−6 to −24 dB
ABLETON
Multiband Dynamics (mode expansion)
MAX
omx.comp~ (ratio inversé) · gen~ custom
gotcha · at extreme ratios, the expander no longer corrects dynamics — it cuts them.
/ 09DYN
TRANSIENT SHAPERlevel-independent envelope
Detects the envelope rather than the level: punches up the attack, opens up or tightens the sustain. Indifferent to playback gain — a quiet snare and a loud one get the same treatment.
ATTACK
−15 to +15 dB
SUSTAIN
−24 to +12 dB
SENS
envelope detector speed
ABLETON
Drum Buss · Transient Shaper (M4L Pack)
MAX
slide~ (env. follower fast/slow) · gen~
gotcha · the transient shaper acts on envelope shape, not on signal level.
/ —NOTE
SIDE-CHAININGnot an effect, a routing
Any compressor, gate, expander, de-esser or dynamic EQ becomes a different beast when its detector listens to a different source. Pumping kicks. Ducking pads. Voice-triggered FX. The trick that built modern dance music.
KEY
any audio bus
FILTER
HP / LP on detector
EXTERNAL
required for ducking
/ 10—11
HARMONIC GENERATION
non-linear processing — adding what wasn't there
/ 10HAR
SATURATIONsoft — tanh, tape, tube
A non-linear transfer function that gently bends the wave — adding low-order harmonics (2nd, 3rd, 5th) without snapping. Adds weight, presence, "glue" and apparent loudness without the brittleness of clipping.
DRIVE
0 – 24 dB pre-gain
TYPE
tape · tube · transformer · tanh
BIAS
even / odd harmonic emphasis
MIX
0 – 100 % (parallel)
ABLETON
Saturator · Roar · Drum Buss · Pedal
MAX
tanh · lookup~ · overdrive~
gotcha · saturation adds harmonics and increases perceived loudness; any comparison demands a level-match.
/ 11HAR
DISTORTIONhard / wave-shape / bit-crush
Hard clipping, wave-shaping, fold-back and quantisation noise. Where saturation flatters, distortion attacks: the signal is no longer the same instrument. Used as colour, as character, or as a destructive design tool.
gotcha · without oversampling, spectral content generated above half the sampling frequency folds back as aliasing.
/ —NOTE
HARMONIC ORDEReven = warm, odd = aggressive
Symmetric clipping (push/pull) generates odd harmonics: 3f, 5f, 7f — the sound of solid-state, transistors, op-amps. Asymmetric clipping (single-ended, transformer) generates even harmonics: 2f, 4f — the sound of tubes and analogue tape. Both have a place.
PAIR
warm, octave-up, "fat"
IMPAIR
aggressive, "edgy", presence
LES DEUX
most real circuits — tilt by bias
/ —NOTE
PARALLEL PROCESSINGdry/wet · M/S · multiband
Most modern engineers process in parallel: heavy distortion, compression or saturation on one path, dry signal on another, blended at the end. You keep the body of the source intact and add the texture next to it instead of over it.
SOMME
phase-aligned mix
NY UPWARD
dry + heavily-compressed copy
M/S
process mid and side independently
/ 12—13
TIME-BASED
delayed reflections, simple and dense
/ 12TIM
DELAYtape · digital · ping-pong · mod
Discrete copies of the input, time-shifted and re-fed. Synced to the grid for rhythmic effects, free for slap-back, modulated for chorus-like motion, filtered in the feedback loop for the classic dub decay.
TIME
0.1 ms – 5 s · synced 1/64 – 4 bars
FB
0 – 100 % (>100 % self-oscillates)
FILTER
HP / LP in feedback path
MIX
0 – 100 % wet
ABLETON
Delay · Echo · Grain Delay · Filter Delay
MAX
delay~ · tapin~/tapout~ · delread~/delwrite~
gotcha · each echo generation inherits the spectral artifacts of the previous one.
/ 13TIM
REVERBconvolution · algorithmic · spring · plate
Reverb is statistically dense delay. Early reflections (5–80 ms) tell the ear about size and material. The diffuse tail tells it about volume and decay. Convolution captures real spaces; algorithms invent impossible ones.
SIZE
0 – 100 % (room → cathedral)
RT60
0.1 – 30 s decay to −60 dB
PRE-DLY
0 – 250 ms (separates source from space)
DAMP
HF rolloff in tail
DIFFUSION
echo density
ABLETON
Reverb (Hybrid) · Convolution Reverb (M4L)
MAX
gigaverb~ · yafr2~ · freeverb~
gotcha · low-frequency energy accumulates in the decay; an HPF on the send contains it.
/ —NOTE
EARLY REFLECTIONSthe part of reverb that locates
The first six to twelve reflections off walls, floor and ceiling are how your ear infers room dimensions. Strip them, you get a synthetic space; emphasise them, you get a tighter sense of place. The pre-delay parameter is the gap between source and these first echoes.
FENÊTRE ER
5 – 80 ms post-source
GAIN ER
relative to tail
MOTIF
determined by virtual geometry
/ —NOTE
DUCKED REVERBside-chain on the wet bus
Send a vocal to a long reverb, then sidechain-compress the reverb return from the dry vocal. The tail blooms in the gaps between phrases, vanishes during them. Cleanest way to drown a voice in space without losing intelligibility.
RATIO
4:1 – 10:1 on wet bus
ATK / REL
fast / slow
KEY
dry voice (pre-FX)
/ 14—19
MODULATION
a parameter as a function of time
/ 14MOD
CHORUSshort modulated delay
Two or more delayed copies (15–35 ms) whose delay times wobble slowly. The detuned, slightly out-of-time copies sum with the dry to create the impression of a section playing one part — width, motion, plurality.
DELAY
10 – 40 ms base
RATE
0.1 – 8 Hz LFO
DEPTH
0 – 10 ms modulation
VOICES
2 – 8 detuned copies
ABLETON
Chorus-Ensemble
MAX
delay~ + cycle~ (LFO) · tapout~ modulé
gotcha · the mono sum of a chorus produces comb filter cancellations.
/ 15MOD
FLANGERswept comb filter
A delay below 15 ms summed back into the input creates a series of evenly-spaced notches in the spectrum (a comb filter). Sweep the delay length and the notches sweep with it — that hollow, jet-engine "whoosh".
DELAY
0.5 – 15 ms base
RATE
0.05 – 5 Hz
DEPTH
0 – 100 %
FB
−95 to +95 % (sign matters)
ABLETON
Phaser-Flanger
MAX
delay~ court + feedback · comb~
gotcha · negative feedback eliminates even-order notches and accentuates the hollow signature.
/ 16MOD
PHASERcascaded all-pass filters
A chain of all-pass filters whose notches are not harmonically spaced (unlike a flanger). Sweep the centre and you get the fluid swirl that defined 70s electric pianos and guitars. No comb structure — a totally different beast from flanging despite the family resemblance.
STAGES
2 / 4 / 6 / 8 / 12
RATE
0.05 – 10 Hz
DEPTH
frequency span of sweep
FB
resonance at the notches
ABLETON
Phaser-Flanger
MAX
allpass~ cascadés · cycle~
gotcha · the phaser modifies the signal's phase without altering its harmonic content.
/ 17MOD
TREMOLOamplitude modulation, sub-audio
An LFO drives a VCA on the signal — pure amplitude modulation at sub-audio rates. Pulses, pulses harder, pulses to silence. Above ~20 Hz it crosses into ring modulation territory and stops being tremolo.
RATE
0.1 – 20 Hz · synced to grid
DEPTH
0 – 100 %
SHAPE
sine · tri · square · saw · S&H
PHASE
L/R offset → vibrato-pan
ABLETON
Auto Pan (modes phase L=R)
MAX
*~ × cycle~ (LFO)
gotcha · beyond 20 Hz, amplitude modulation enters the audible zone and becomes ring modulation.
/ 18MOD · UTL
AUTO PANtremolo split L/R, 180° apart
Two tremolos in stereo, ninety or one-eighty degrees apart, applied to the L and R channels separately. The signal sweeps across the stereo field on a periodic cycle — pure motion, no spectral colour.
RATE
0.05 – 10 Hz · synced
WIDTH
0 – 100 % L/R amplitude
OFFSET
L/R phase, 90° = circular
SHAPE
sine, tri, S&H, random
ABLETON
Auto Pan
MAX
pan~ · *~ × LFO L/R
gotcha · the stereo movement created by an auto pan disappears in the mono sum.
/ 19MOD · HAR
RING MODULATORmultiplication of two signals
Multiply audio by a sine carrier. The output contains the sum and difference frequencies of every spectral component — usually inharmonic. Bell-like, metallic, alien. Indispensable for sound design, near-useless for mixing.
FREQ
1 Hz – 5 kHz carrier
SOURCE
sine · external audio
MIX
nearly always < 50 %
ABLETON
Frequency Shifter (mode Ring)
MAX
*~ (deux signaux) · cycle~ * input~
gotcha · multiplying two signals generates frequencies absent from both original signals.
PSOLA, phase-vocoder, or formant-corrected: change the pitch without changing the playback duration. Useful in semitones for harmonisation, in cents for de-tuning, and in micro-tones for character (hello, formant-shifted vocals).
RANGE
−24 to +24 semitones
FINE
±100 cents
FORMANT
preserved · shifted independently
ALG
granular · phase vocoder · PSOLA
ABLETON
Frequency Shifter · Pitch Shifter (M4L) · Grain Delay
MAX
gizmo~ · pfft~ + phase vocoder · grain~
gotcha · pitch displacement without formant preservation alters the timbral identity of a voice.
/ 21PIT
TIME STRETCHduration translation, pitch-preserved
Inverse problem: change duration, keep pitch. Granular synthesis (slice, re-trigger, overlap) handles polyphonic material reasonably; phase-vocoder handles harmonic material cleanly; transient-aware modes preserve the snap of drums.
gotcha · no time-stretch algorithm is transparent; all leave a signature at extreme ratios.
/ 22
STEREO / IMAGING
manipulating perceived width
/ 22UTL
STEREO WIDENERM/S re-balance · Haas · all-pass
Three families: M/S processing (boost the side channel relative to the mid), Haas-effect (5–35 ms inter-channel delay), and all-pass phase rotation. Each one widens differently and trades mono compatibility for image size.
WIDTH
0 – 200 % (M/S gain)
HAAS
5 – 35 ms inter-channel delay
RANGE
frequency-dependent (low = mono safe)
MONO ≤
120 Hz typically
ABLETON
Utility (Width) · Pedal (mode stereo)
MAX
+~ / -~ (encodage M/S) · pan~
gotcha · what opens the stereo cancels in mono in proportion.
/ —NOTE
MID / SIDEnot an effect — a coordinate system
Sum and difference of the L/R pair. Mid is the mono content (vocal, kick, bass, snare); Side is everything that differs (room, reverb, hat panning). Process them independently, sum back, gain a control axis the L/R worldview can't give you.
ENCODE
M = (L+R)/√2 · S = (L−R)/√2
DECODE
L = M+S · R = M−S
USAGE
EQ, compression, saturation per channel
ABLETON
Utility · M/S Tools (M4L)
MAX
matrix~ · +~ / -~ · pan~
/ —NOTE
UTILITY GAINthe most under-used plugin
Level-match. Gain-stage. Re-balance after a non-linear plugin pushed everything 4 dB louder. Phase-flip a duplicate to find the difference between two takes. The boring plugin that fixes more problems than the famous ones.
GAIN
−∞ to +24 dB
POLARITY
+ / −
L/R BAL
independent gains
WIDTH
M/S gain ratio (when present)
ABLETON
Utility
MAX
gain~ · *~ · live.gain~
/ —NOTE
METERINGpeak · RMS · LUFS · true-peak
Peak shows transients (will it clip?). RMS shows energy (will it sound loud?). LUFS shows perceived loudness (will streaming services normalise it?). True-peak shows what the DAC will actually try to output. Use all four. Trust LUFS for translation.
PEAK
sample-accurate, fast
RMS
300 ms window, energy
LUFS-I
−14 streaming · −16 podcast · −23 broadcast
TRUE PK
oversampled, dBTP
ABLETON
Spectrum · VU Meter (M4L) · LUFS Meter (M4L)
MAX
meter~ · snapshot~ · peakamp~
SIGNAL CHAIN — INSERT ORDER ON A TYPICAL CHANNEL→ time flows left to right
Compression-then-EQ and EQ-then-compression do not produce the same sound. Either is correct; neither is universal. The chain is part of the instrument.
// 02 — gain stage
Every non-linear plugin (saturation, compression, distortion) changes loudness. Level-match A/B at every step or you will mistake "louder" for "better" — every time.
// 03 — bypass to compare
If you can't hear what the plugin does at this setting, the plugin doesn't do anything at this setting. Bypass and second-guess.
// 04 — subtractive first
Cut what's wrong before adding what's missing. EQ-cut, then EQ-boost. Gate, then compress. The sound was already there before you started.
// 05 — context wins
The vocal that sounded perfect solo will disappear in the mix. The bass that thumps in headphones will boom in the car. Always check in context.
// 06 — restraint is technique
Six plugins doing 1 dB each beats one plugin doing 6 dB. The art is in the gentle stack, not the heroic correction. Less knob, more ear.
Going further
Structure Void training
What you just read, we practise it in session. Small groups, 100% adapted to your project, French-eligible funding (AFDAS, OPCO, employer).
Structure Void is Qualiopi-certified for its training, and is an Ableton Certified Training Center. Julien Bayle is an Ableton + Max Certified Trainer.
REFERENCE is a series of technical cards compiled in the studio. For artists, sound designers, engineers and educators who want dry, sourced, signal-not-noise content.
This first card covers audio effects. Future ones will cover synthesis, MIDI, gesture-to-sound, RNBO, and everything that patches.