🎛️ Built-in Audio Effects and DSP Processors

Omni DAW comes with a suite of 34 built-in DSP processors designed for sound shaping, correction, compression, spatialization, and creative audio processing. Effects are connected in an Insert Chain on each track, the Master bus, and each Return bus — every one of them has this chain by default, so it doesn't need to be "added" separately. In addition to native effects, external VST3, CLAP, LV2, and (legacy) VST2 plugins can be loaded into the same chain, as well as custom effects written in OmniScript (.omfx, @mode: audio — see /docs/omniscript).

1. EQ (Parametric Equalizer)

A 16-band mastering-grade equalizer with a real-time Spectrum Analyzer (post-EQ fill, pre-EQ line).

• Band shapes (8): Bell, Low Shelf, High Shelf, Tilt Shelf, Low Cut, High Cut, Band Pass, Notch. Cuts offer slopes of 6, 12, 18, 24, 36, 48, 72 and 96 dB/oct.
• Per-band channel: Stereo, Mid, Side, Left or Right — e.g. mono the low end with a Side low cut, or add air only to the sides.
• Zero Latency mode: analog-matched filters that keep their true analog shape right up to Nyquist (no "cramping" of high bells and shelves), processed in 64-bit, with smoothed parameters and click-free shape changes.
• Linear Phase mode (Low / Medium / High resolution): no phase shift at all; its latency is compensated automatically (PDC).
• Global: Output, Gain Scale (0–200 % of every band gain), Auto Gain (keeps loudness constant while you shape), A/B compare.
• Editing: drag a handle for frequency/gain (Shift = fine), mouse wheel for Q, double-click empty space to add a band, double-click a handle to reset it, Alt-click to switch a band on/off, right-click for the band menu, hold the headphones key to listen to a band on its own. The expand button opens a large EQ window.
• Presets: a factory library of mastering curves (Mastering, Mastering M/S, Utility) plus your own presets saved as .eqomni files; the preset menu also imports and exports .eqomni files.

2. Filter (Multi-model Character Filter)

A character filter with 12 models, a built-in LFO and envelope follower and a live view of where the cutoff actually is. The curve on screen is computed by the same code that processes the audio, so what you see is what you hear.

• Models (menu in the header): Clean transparent zero-delay-feedback state-variable filter; Ladder and Diode the same core with a soft-saturating (Diode: asymmetric) resonance; Moog a transistor ladder with a nonlinear pole in every stage, so loud signals and resonance compress the way the hardware does; Korg 35 the MS-20 Sallen-Key with its clipper in the loop; OB-Xd the Oberheim SEM state-variable (Blend leans it toward band-pass); Vintage a ladder with linear poles; Comb a resonant comb tuned by the cutoff (Neg flips its polarity); Cut Warp / Reso Warp let the input level push the cutoff or damp the resonance (Warp); S&H a sample-and-hold decimator; Vowel five formant filters that morph A-E-I-O-U.
• Mode and slope (keys in the header): LP, HP, BP, Notch, AP, Peak and Morph, which sweeps continuously LP → BP → HP → Notch with the Morph knob; 12, 24 or 48 dB/oct. Changing mode glides instead of clicking; changing model, slope or oversampling fades the filter out and back in within a few milliseconds.
• Character: Drive pushes the signal into the filter, FB adds saturation inside the resonance loop, Curve picks the saturation shape (Tanh, Asym, Soft, Diode, Fold, Hyst), oversampling 2× or 4× removes aliasing from the nonlinear models (with automatic delay compensation).
• LFO: Amount (± octaves), Rate in Hz or, with Sync, a note division from 1/32 to 8 bars; synced LFOs follow the song position, so a sweep always lands on the same beat. Seven shapes (Sine, Triangle, Saw Up/Down, Square, S&H, Drift) and Stereo phase up to 180° between left and right.
• Envelope: the input level opens (positive Amount) or closes (negative) the filter, with Attack, Release and Sens. On a track, the SC picker keys the envelope from another track or return bus, for example a kick that makes a pad breathe.
• Spread offsets the left and right cutoff by up to ±2 octaves for width.
• Output: Output trim, Mix (dry path time-aligned, also with oversampling) and Auto gain, which takes back half of the Drive and restores the bass the Moog/Vintage ladders lose at high resonance.
• Display: drag anywhere on the response to move the cutoff (sideways) and resonance (up/down), Shift for fine moves, wheel for resonance, Ctrl + wheel for Morph/Vowel, double-click to reset. With LIVE on you see the spectrum, the range the modulation can reach (shaded, with its two extreme curves) and the modulated curve of each channel in real time. The motion view next to it scrolls the actual cutoff of both channels and the envelope level over 2, 4 or 8 s, above a preview of the LFO shape and IN/OUT meters.
• Presets: factory library (Essentials, Analog Character, Movement, Dynamic, Creative) plus your own .filtomni files, import/export and A/B compare.

3. Compressor (Dynamics Compressor)

A mastering-grade compressor with six characters and live metering straight from the engine. Every style is a different circuit, not a label: the curve you see is computed by the same code that processes the audio.

• Styles (keys in the header): CLEAN transparent feed-forward digital compression; VCA punchy decoupled envelope with a firm knee; GLUE bus compressor with a soft knee and program-dependent release; OPTO optical feedback leveler (fast first release, long glowing tail that grows the longer it compresses); FET ultra-fast feedback limiter whose colour rises with the gain reduction; V-MU variable-mu tube, the ratio grows with level. Opto and FET are true feedback designs solved without the usual one-sample delay, so they stay stable at any ratio and attack. Switching styles morphs smoothly, without clicks.
• Dynamics: Thresh, Ratio (1:1 to ∞:1), Knee (0–24 dB) and Range (the maximum gain reduction, useful for gentle "never more than 3 dB" mastering moves).
• Timing: Attack (0.01 ms – 250 ms), Release (5 ms – 5 s), Hold (keeps the reduction before the release starts) and Look (lookahead up to 20 ms: the gain moves before the transient arrives; the latency is compensated automatically). AUTO REL makes the release program-dependent: fast after isolated transients, slow under sustained compression.
• Detector: Pk/RMS blends continuously from peak to RMS detection, Link sets the stereo link, SC HP / SC LP filter the key signal (for example so the kick does not pump a mix bus) and KEY lets you listen to the filtered key. The channel keys choose ST (left/right), M/S (mid and side compressed separately), MID only or SIDE only. On a track, the SC picker keys the compressor from another track or return bus.
• Output: Color (anti-aliased harmonic saturation of the gain element, voiced per style and rising with the gain reduction; at 0 the path is bit-transparent), Makeup, AUTO GAIN and Mix for parallel compression.

The dynamics history scrolls the input level (steel), the output level (green) and the gain reduction (red, hanging from the top on an auto-scaling axis) on a true time axis, with the threshold and knee drawn across. Drag it vertically to move the threshold, click the time chip to change the window (2–16 s), double-click to clear. The transfer curve shows the static input/output curve with the reduction wedge; the live ball sits at the detector level with the reduction actually applied, and its fading trail shows the attack and release at work. Drag the threshold node sideways, the ratio node at the right edge up and down, and use the mouse wheel for the knee. The meter bridge shows input, gain reduction per channel (L/R or M/S) and output with peak holds (click to reset).

Presets: a factory library (Vocal, Bus, Drums, Instrument, Mastering and Creative categories) plus your own .compomni files in the compressor_presets folder of the Omni configuration directory, with import/export and A/B comparison. Projects saved with the earlier Compressor load with their settings unchanged.

4. Gate (Noise Gate)

A gate, downward expander and ducker with live metering straight from the engine. The curves on screen are computed by the same code that processes the audio.

• Mode (keys in the header): GATE closes the signal when the key falls below the threshold; DUCK turns it around, so the key pushes the signal down while it is above the threshold (music under a voice-over, a bass under the kick).
• Gate: Thresh, Hyst (hysteresis: an open gate closes only below Threshold − Hysteresis, so a decaying note does not chatter around the threshold), Range (how far the closed gate attenuates, down to a true mute at −∞), Ratio (∞ is a hard gate, lower values a gentle downward expander) and Knee.
• Time: Attack is the 10–90 % opening time (from 0.01 ms), Hold keeps the gate open after the key has gone, Release is the time the gain needs to fall 60 dB (an RT60 of the gate, up to 10 s) and Look is a lookahead of up to 20 ms, so the gate is already open when the transient arrives (the latency is compensated automatically).
• Envelope styles: CLASSIC a first-order envelope like a hardware VCA gate, the tightest and punchiest; SMOOTH a second-order envelope without corners, the cleanest on vocals, guitars and pads; ADAPT a program-dependent release, fast after short hits and longer after sustained notes. The times mean the same in every style.
• Key: the detector holds each peak for half a period of the lowest frequency the key carries, so low notes read as a steady level and do not open and close the gate on every cycle. Pk/RMS blends to RMS detection, Link sets the stereo link, HP and LP on the strip under the history switch the key filters (Key HP / Key LP set their cutoffs) for frequency-conscious gating, and KEY lets you listen to the filtered key. On a track, the SC picker keys the gate from another track or return bus.
• Output: Mix (parallel gating) and Output (±24 dB).

The gate history scrolls the input (steel), the output (green, blue when ducking) and the key level (cyan) on a true time axis, with the open and close thresholds and the hysteresis band between them. Wherever the gate attenuates, a dark curtain falls over the history, and the ribbon at the bottom shows the state: open, hold, closing or closed. Drag the amber line to set the threshold and the copper line to set the hysteresis; click the time chip to change the window (2–16 s), double-click to clear. The transfer display draws both branches of the curve, the opening path and the closing path, so the hysteresis shows as a loop; the live ball sits at the key level with the gain actually applied. Drag the threshold, close and range nodes; the mouse wheel sets the ratio, Shift + wheel the knee. The envelope display shows the gain of a key burst through the real envelope: the opening zoomed in on the left (with the point where the delayed audio arrives when lookahead is on), hold and release on the right, the other two styles as ghosts. Drag the A, H and R handles. The meter bridge shows the state lamp, input, gain and output with peak holds.

Presets: a factory library (Essentials, Drums, Voice, Instruments and Creative) plus your own .gateomni files in the gate_presets folder of the Omni configuration directory, with import/export and A/B comparison. Projects saved with the earlier Noise Gate load with the same sound: their attack and release times are converted to the new definitions.

5. Delay (Stereo Delay)

A studio delay with five engines, tempo sync and a live echo display. Everything the repeats go through (filters, diffusion, saturation) sits inside the feedback loop, so every repeat is a little darker, softer or more smeared than the one before, as on real hardware.

• Modes (keys in the header): DIG pristine digital repeats (bit-exact below 0 dBFS); BBD a bucket-brigade line with a 2:1 compander that breathes with the signal, clock noise and a bandwidth that falls as the time grows (long analog echoes are genuinely dark); TAPE saturation with even harmonics, wow and flutter, playback-head bump and hiss; MULTI four panned heads on one loop, feedback from head 4 (a Space Echo style rhythm delay); LCR left, centre and right taps with feedback from the centre only. In MULTI and LCR the DIG / BBD / TAPE keys choose the colour of the heads. Switching modes is click-free.
• Time: Time in ms (1 ms to 4 s) or a tempo-synced division when SYNC is on (1/64 to 2 bars, including dotted and triplet values; switching sync on or off keeps the echo where it is). Offset delays the right side against the left (±50 %) for wide stereo patterns. FADE / GLIDE chooses what a time change does: a clean cross-fade, or a tape-style pitch glide with its own glide time.
• Routing: STEREO (each side echoes itself), PING (the sum bounces left, right, left…), CROSS (stereo in, every repeat swaps sides).
• Loop: Fdbk up to 110 % (above 100 % the loop self-oscillates but stays bounded by a soft ceiling), Lo Cut (12 dB/oct) and Hi Cut (6 dB/oct) in the loop (at their end stops they are off), Diffuse smears each repeat through an allpass diffuser into a reverb-like cloud without losing the beat.
• Character: Drive (anti-aliased saturation, level-compensated), Noise (BBD / tape hiss), Rate and Depth / Wow (modulation of the read heads: stereo chorus on DIG/BBD, mono wow with drift and flutter on TAPE).
• Output: Duck lowers the echoes while the input plays (threshold and release on the key strip) so a vocal stays clear and the repeats bloom in the gaps; Width (0–200 %), Mix and Output (±12 dB). FREEZE (snowflake) holds the loop forever and ignores new input.

The echo field draws the echo train the current settings produce: the left channel upwards, the right downwards, the dry hit at the left edge. Each repeat is a stem whose height is its level and whose colour dulls as the loop filters take its top end away; beat lines follow the project tempo and a scan line sweeps at real-time speed, lighting each repeat as it would play. Drag horizontally in the upper half to set Time (in sync mode it steps through the divisions), in the lower half to set Offset, and vertically to set Feedback. In MULTI and LCR, grab a head: sideways sets its time (snapped to 1/16 notes when synced), up/down its level. The LIVE view scrolls the input (steel), the echoes the engine actually produced (mode colour) and the ducking (red) on a true time axis. LOOP TONE shows one pass through the loop filters and what is left after 2, 4 and 8 repeats; drag its handles to set Lo / Hi Cut, double-click one to switch it off.

Presets: a factory library (Basics, Rhythmic, Vintage, Space and Creative) plus your own .delayomni files in the delay_presets folder of the Omni configuration directory, with import/export and A/B comparison. Projects saved with the earlier Delay load with their settings; the old ping-pong switch now gives a true ping-pong (the old channel-swap behaviour is CROSS).

6. Reverb (Algorithmic Reverb)

An algorithmic space generator (FDN network). Room Size scales the actual geometry (delay lines and early reflections), while Decay sets the RT60 independently, so a small room can ring long and a hall can be short. Damping controls the HF absorption, Pre-Delay the gap before the tail, Mod Depth / Mod Rate the tap modulation that keeps the tail from sounding metallic, and Width / Mix the output. Freeze holds the tail indefinitely. The effect panel includes an Algorithmic / Convolution switch, which replaces the engine with the Convolution Reverb (see below).

7. Convolution Reverb

A reverb based on impulse response files (IR, up to approx. 8 seconds at 48 kHz), processed using partitioned FFT convolution. It allows for replicating the acoustics of real rooms — accessible from the Reverb effect panel.

8. Chorus (Ensemble Chorus)

One delay line, three modes, picked with the CHORUS / DIM / FLANGE keys. Switching fades the wet signal out and back in over 8 ms, so it never clicks. Every value is kept; only a Delay outside the new mode's working range is moved into it (a 12 ms flanger would be a slow echo).

• CHORUS: a multi-voice stereo ensemble chorus. Each channel sums up to 4 modulated delay taps (Voices). Every voice sweeps around its own staggered centre delay with its own slightly detuned LFO, which spreads the comb notches across the spectrum. The voices are loudness-normalized, so adding voices makes the sound richer rather than louder.
• DIMENSION: two lines, left and right, swept by the same LFO in antiphase, each channel taking the other line inverted. In stereo it is wide; summed to mono the two sweeps cancel, so there is no pitch wobble and nothing to lose in mono (the two-line rack ensemble of the 80s). The I–IV switches set Rate, Depth and Delay like the unit's buttons; Voices, Spread and the feedback loop are not used.
• FLANGER: one short tap per channel with the whole feedback loop. Delay goes down to 0.1 ms for the jet sweep, INV inverts the feedback (hollow notches instead of ringing peaks), Spread offsets the right channel's sweep.

The controls, laid out for the mode:

• Motion: Rate (Hz) or a tempo-synced division when SYNC is on (8 bars down to 1/16, including dotted and triplet values). Depth is the sweep width in ms, and Delay is the centre of the sweep. The four LFO shapes are SIN, TRI, RND (smooth random wander) and ENS (the two-speed ensemble LFO of classic string machines).
• Ensemble (CHORUS): Voices (continuous, voices fade in and out without clicks), Spread (L/R LFO phase offset, with odd voices mirrored so the ensemble fills the whole field) and Width (0–200 % mid/side width of the wet signal). DIMENSION shows its switches and Width here, FLANGER Spread and Width.
• Tone: Lo Cut and Hi Cut (12 dB/oct on the wet signal). Vintage adds bucket-brigade character: an anti-aliased (ADAA) input saturation, the BBD's reconstruction low-pass and a slow random clock drift.
• Loop (CHORUS, FLANGER): Feedback, Damp (darkens the loop) and INV (feedback polarity).
• Output: Mix and Output gain (±12 dB).

The panel shows the voices live. In the voice field, X is delay time and the centre line is "now". The left channel's history scrolls up and the right channel's scrolls down, so you see the real LFO waveforms, the stereo spread and the feedback repeats (hollow ghosts). Drag the field horizontally to set Delay and vertically to set Depth. Next to it, the stereo image shows where each voice sits and how wide the wet signal is, and wet tone shows the filter curve (dashed: feedback damping). In FLANGER that display becomes the comb: the response of dry plus the looped tap at the delay the left channel reads right now (the right channel's faintly), computed by the same formula as the engine's loop.

Presets: a factory library (Classic, Ensemble, Dimension, Instrument, Flanger and Creative categories) plus your own presets saved as .chorusomni files in the chorus_presets folder of the Omni configuration directory. Use the arrows to step through them, the folder button to browse, import and export, and the save button to store the current sound. A/B compares two settings. Projects saved with the earlier Chorus load with their original sound.

9. Phaser (Multi-stage Phaser)

A phaser based on an all-pass filter chain (2–12 stages), with Depth, Feedback, and Mix controls.

10. Saturator (Harmonic Saturation)

A non-linear saturation processor that adds harmonics and "warmth" to the signal.

11. Multiband Dynamics (Multiband Compressor)

A 3-band compressor with frequency splitting via 4th-order Linkwitz-Riley filters (24 dB/oct) for precise dynamics control across the low, mid, and high bands.

12. Modal Resonator (Resonator Bank)

A physical-modelling resonator in the spirit of Mutable Instruments Rings: the input strikes a bank of up to 32 resonant modes per voice, turning drums, noise, vocals or any sound into a ringing string, bar, drum skin, bell or pipe. The modes are modulation-safe filters, so every knob can be swept and automated without clicks, and switching models morphs the ringing tail instead of cutting it.

• Models (keys under the body display): STR modal string (Structure = stiffness, from harmonic to piano-like stretched partials), SYM sympathetic strings (Structure picks the chord: Unison, Overtones, Fifths, Major, Major 7, Minor, Minor 7, Dominant 7, Sus4, Add9, Cluster), INH inharmonic metal, FM FM sideband spectrum, QNT partials snapped to semitones, RVB long detuned reverb strings, BAR free bar to tuned marimba (1 : 4 : 10), MEM drum skin (ideal membrane to tuned drum), BELL minor-third to major-third church bell, TUBE closed pipe (odd harmonics) to open pipe.
• Tune: Freq (MIDI note, C-1 to C7), Fine (±100 cents) and Struct.
• Body: Bright (dark bodies lose their upper partials and let them die faster, bright bodies ring evenly), Decay (30 ms to 30 s; the level stays the same when you lengthen the decay, only the tail grows), Pos (strike position, with the classic comb character of striking a node) and Modes (1–32 partials, cross-faded).
• Exciter: Drive (±24 dB into the resonator), Lo Cut and Hi Cut (12 dB/oct on the signal that strikes the modes; a lower Hi Cut acts like a softer mallet).
• Output: Spread (the fundamental stays centred and mono-safe, the upper modes fan out left and right), Mix and Output (−24 to +12 dB). A transparent safety limiter keeps the wet signal below −1 dBFS when a sustained tone lands exactly on a mode.
• MIDI: with MIDI on, the track's notes tune the resonator. Keys 1–4 set the polyphony: every held note owns a voice and receives the input, and a released voice rings out naturally, so you can play chords of strings or bells over any audio.

The mode spectrum shows every mode the engine rings on a 20 Hz – 20 kHz axis: height is the mode's level, the soft aura grows with its decay time, and hovering a mode shows its frequency, note, T60 and level. With LIVE on, the output spectrum glows underneath and each mode lights up with the energy it is ringing right now. Drag horizontally to tune, vertically to set the decay, and use the mouse wheel for Structure. The body display animates the struck object (string, bar, pipe, drum skin, bell or the set of sympathetic strings) from the same modes. Click or drag it to move the strike position.

Presets: a factory library (Strings, Harmony, Percussion, Metal, Space and MIDI categories) plus your own .resonatoromni files in the resonator_presets folder of the Omni configuration directory, with import/export and A/B comparison. Projects saved with the earlier Modal Resonator keep their settings (with the old 24-mode count) and load on the new engine.

13. Granular Clouds (Textural Granulator)

A granular processor that records the input into a stereo buffer (up to 8 seconds) and plays it back as a cloud of up to 64 grains: ambient washes, shimmering pitch halos, time-stretched smears, frozen pads, rhythmic stutters and transient-triggered bursts. Grains read the buffer through anti-aliased copies of the audio, so transposing up stays clean instead of folding high frequencies back down, and each grain is placed so that it never reads audio that is being overwritten, at any pitch or direction: no clicks from the record head.

• Clock (keys in the header): FREE a continuous cloud: Density sets how many grains overlap (from isolated grains to a wash of 32) and Chaos moves the timing from metronomic to random. SYNC fires grains on the song grid at a division (1/64 to 1 bar, including dotted and triplet values); Density becomes Ratchet (1 to 4 grains per tick) and Chaos humanises and drops ticks. ONSET fires a burst on every transient of the input; Density becomes Burst (1 to 8 grains) and Sens sets the detector.
• Grain: Position picks where in the buffer grains read (right = newest), Size (1 ms to 4 s), Pitch (±24 semitones), Texture morphs the grain window from a box (clicky, rhythmic) through Tukey and Hann to a narrow bell, and Skew moves its peak to the start (percussive) or the end (swelling, reverse-like).
• Scatter: Spray scatters grains around the position, Vary randomises their length, Detune their pitch (up to ±6 semitones) and Reverse is the chance that a grain plays backwards. The scale menu snaps every grain's pitch to Chromatic, Major, Minor, Penta, Minor Penta, Fifths or Octaves, so Detune produces harmonic clouds instead of noise.
• Motion: Scan moves the read point through the buffer: 0 follows the input, −1 holds the read point still while the input keeps recording (time stretch), −2 travels backwards, positive values race ahead; when frozen, +1 plays the captured audio forward in real time. Buffer sets how much history the grains can reach (0.5 to 8 s).
• Space: Spread places grains across the stereo field, Width (0 to 200 %), Diffuse smears the cloud through an allpass diffuser, Reverb adds the plate reverb with its own Decay and Damp.
• Output: Fdbk records the cloud back into the buffer (regenerating textures; a soft ceiling keeps it stable even at 100 %), Lo Cut and Hi Cut filter the wet signal and the feedback loop (the loop darkens or thins as it regenerates), Duck lowers the cloud while the input plays, Mix (equal-power) and Output. The cloud's loudness is normalised as it thickens, so higher densities sound fuller, not louder.
• FREEZE (snowflake) stops recording: the grains keep playing what is in the buffer. A frozen buffer survives transport start and stop.

The cloud field shows the record buffer as a waveform (oldest left, the write head pulsing on the right edge, ice-blue when frozen) and every grain in flight as a glowing particle: across at the moment it is reading, down by its pan (left at the top), its colour from its pitch (violet below, mint at unison, gold above, pink when reversed), its size from its window and a tail in the direction it travels, with a short afterglow of where grains have just been. Click or drag horizontally to set Position, vertically for Spread; the wheel sets Spray (Ctrl + wheel: Size). The grain lens draws the window, the train of neighbouring grains the clock produces and their summed power (flat means a smooth cloud, ripples an audible grain rate), with the live grains riding the curve; drag sideways for Skew and up/down for Texture. Pitch space plots every grain by pitch and pan over the octave lines and the selected scale, with the jitter band and IN / WET / FB meters; drag up/down to set Pitch (snapped to the scale, Shift for free), sideways for Spread, wheel for Detune. Without LIVE (or inside Sequenced FX) the displays show a preview cloud that follows the knobs.

Presets: a factory library (Textures, Shimmer, Stretch & Freeze, Rhythm, Glitch, Mix) plus your own .cloudsomni files in the clouds_presets folder of the Omni configuration directory, with import/export and A/B comparison. Projects saved with the earlier Granular Clouds load with their settings, including the grain density and window they had.

14. Transient Shaper

Shapes the hit (Attack) and the decay (Sustain) independently of the signal level, with a live view of single hits. The curves on screen are computed by the same code that processes the audio.

• Shape: Attack (±24 dB) lifts or softens the start of every hit, Sustain (±24 dB) lengthens or shortens its decay (tail, room).
• Times: Atk T is the attack window, how long after an onset Attack acts; Sus T is the sustain window, after how long a falling level counts as the tail.
• Detector styles (header keys): PUNCH compares a fast and a lagging envelope in decibels, so it reacts to the shape of a hit rather than its loudness (a soft and a loud hit get the same treatment), with an instant onset; SMOOTH is the same detector with a softened onset, an S-shaped response and a second-order gain glide, clean on bass, buses and acoustic sources; CLASSIC is the original Omni detector (ratio of a fast and a slow envelope), which old projects open with. The detector holds peaks, so the dense micro-peaks of noise, snare wires or a room are never read as new hits, and a new hit after a loud one starts its own decay.
• Level reading: TRUE (default) is the true envelope from the analytic signal (a pair of all-pass filters 90° apart): instant on onsets and ripple-free on low notes, so the shaper never modulates the bass; PEAK reads the rectified waveform (fastest, ripples on bass); RMS a 2 ms average (gentlest).
• Detection: Sens (sensitivity; at 50 % a 6 dB difference drives the full effect), Thresh (hits quieter than the threshold pass untouched, e.g. hi-hat bleed in a snare mic; off at the bottom of the range), Link (stereo link) and Look (lookahead up to 10 ms, so the attack boost lands exactly on the start of the hit; the delay is compensated automatically). On a track, SC lets another track or a return bus drive the detection.
• Focus: Low and High limit the band that is detected and shaped (e.g. only the beater click above 1.5 kHz, leaving the low end of the kick alone). The split is complementary: at zero gain the output is the input bit for bit, and nothing outside the band changes. The FOCUS strip under the history draws this band with the engine's own filter curve; drag its edges, double-click to reset to full range.
• Channels (strip under the lens): L/R, MID (centre only: kick, snare, vocal), SIDE (sides only: room and width) or M/S (two independent detectors).
• Output: Mix (parallel shaping), Output (±24 dB), CLIP (anti-aliased ADAA soft clipper at 0 dBFS, transparent below −3 dBFS, catches the overs of a big attack boost) and the Δ (only what the shaper adds or removes) and BAND (the focus band the detector hears) listen keys.

The history scrolls the input (steel) and output (green) envelopes on a true time axis like a waveform, with the energy the shaper added glowing amber and the energy it removed in blue; under it the gain lane around 0 dB and the detected attack and sustain ribbons. With the threshold on, the detector level (cyan) and the gold threshold lines appear; drag the lines to set it. Click the time chip to change the window (1–8 s), double-click to clear. The hit lens shows one hit magnified on a time axis where the first milliseconds get as much room as the whole tail, levels relative to the hit's peak: the newest hit solid, the previous ones as ghosts (hovering freezes the picture). Without audio the lens shows the engine processing a synthetic hit (KICK, SNARE or PLUCK, picked in the corner). The amber handle sets Attack (↔ attack time, ↕ gain), the blue one Sustain. The meter bridge has ATK and SUS lamps, input, a two-way gain bar and output with peak hold.

Presets: a factory library (Essentials, Drums, Instruments, Mix and Creative) and your own .tsomni files in the transient_presets folder of the Omni config directory, with import, export and A/B compare. Projects saved with the earlier Transient Shaper open in the CLASSIC style with PEAK reading and sound the same.

15. Limiter (Brickwall Limiter)

A lookahead brickwall limiter with true-peak detection: the output never goes over the Ceiling, also between samples. Typically the last device on the Master.

• Styles: TRANS (clean, exact), MODRN (loud, fast multi-stage release), PUNCH (lets the attack breathe), VNTG (soft knee, gentle warmth), GLUE (wide knee, slow release for buses) and SAFE (extra true-peak margin for the final export).
• Ceiling and Knee; Release (or Auto release), Lookahead and Hold; Link (how much the two channels share one gain); True Peak detection; In Gain drives the limiter; Dither adds TPDF dither for a 16-bit bounce.
• M/S: the two detectors, envelopes and gains belong to mid and side instead of left and right, so a centred kick no longer ducks the stereo width. The ceiling still holds on left and right: both components take whatever extra gain keeps the mid peak plus the side peak under it.
• Loudness: the output's loudness (BS.1770: momentary, short-term, integrated; a new measurement starts with every playback) against a Target (−30 to −5 LUFS). Press Learn, play the loudest part of the song: the limiter sets In Gain step by step (one step per 3 s window) until the short-term loudness sits on the Target, then switches Learn off. Match (gain match) turns the output down by what the limiter added, so switching it in and out compares sound, not level. It only turns down, and the printed level is lower while it is on; switch it off before exporting.

The panel shows the level history with the gain reduction, the ceiling curve, the true-peak readout with an overs counter, the IN · GR · OUT meter bridge (GR as M / S in M/S mode) and the loudness column: momentary as a bar, short-term as a line with its maximum, the input's short-term dashed, the Target in amber (drag it vertically), the integrated loudness and its distance to the Target on top, Learn's progress and the match below.

Presets: Mastering (including Wide Master M/S and Loud Master -9 LUFS), Mix Bus, Broadcast & Streaming (including Streaming -14 LUFS, EBU R128 -23 LUFS, Podcast -16 LUFS) and Creative, plus your own. Projects saved with the earlier Limiter sound the same.

16. Sequenced FX

An Effectrix-style effect sequencer: up to 8 audio rows and 8 MIDI rows, each with its own effect switched on and off by the steps of the grid, sample-accurately and independently of the buffer size (a bounce or freeze sounds exactly like the playback). The latency of the row effects is compensated, so the dry/wet crossfades never comb-filter.

• Rows (+ ROW): sequencer units — Repeat (stutter: the start of the step loops, with pitch and decay per repeat, Duty, size from the step to a bar), Reverse (the audio before the step played backwards), Tape Stop (the tape slows to a stop or spins up, with a curve), Scratch (rocks back and forth like a hand on a record), Crush (bit depth and sample-rate reduction) and Volume (gain, pan and width — a trance gate at −∞); any built-in effect (Filter, Delay, Reverb, Chorus, Phaser, Saturator, EQ, Compressor, Limiter, Noise Gate, Transient Shaper, Modal Resonator, Granular Clouds, Multiband), an OmniScript effect and — on tracks — MIDI effects acting on the notes before the instrument (arpeggiator, chord, scale…). The buffer units record their input all the time, so they always have the audio from before the trigger.
• Steps: every step has a level (how much effect), a chance (the probability it plays — rolled from the row's seed, so a bounce is identical to the playback), a gate length, ratchets (1–8 triggers inside the step: rolls and stutters), tie (legato from the previous step: the effect continues instead of restarting) and a lane value. The mode keys in the header (STEP, LEVEL, CHANCE, LENGTH, RATCHET, LANE, TIE) choose what the mouse draws: in STEP the left button switches steps on, the right one off, the wheel sets the level; in the value modes drag vertically to set the value, right-click restores the default.
• Row (the inspector on the right; click a row to select it): edit the effect (its full panel in a window, also by double-click), its own step count and step length (polymeter and polyrhythm), the play order (forward, reverse, ping-pong, pendulum, random, drunk), the row Mix (automatable), the gate Attack and Release (0.3 ms – 2 s, a smooth curve), the tail mode CUT (the effect's output is gated — its tail stops with the step) or RING (its input is gated — delay and reverb tails ring out), RETRIG, the lane (one parameter of the effect set on every step, with a range; logarithmic for frequencies and times), shift, randomize, invert, copy and paste steps, Euclidean rhythms, a new dice seed, row order and removal; M and S next to the row name mute and solo it with a smooth fade.
• Device: step count and step length, Swing, Chance (scales every step's chance: 0 % nothing, 200 % doubles), Mix, Output, rows SER (in series) or PAR (in parallel, aligned to the slowest), clock SYNC (with the transport) or FREE (the steps run at the song tempo even with the transport stopped). All of them, and the Mix of every row, can be automated and modulated.

The grid shows all rows on a shared time axis: each row's first cycle starts at the left edge and its repeats fill the view as dimmed ghosts, so rows of different lengths and speeds read at a glance. A cell shows everything at once — the bar height is the level, its opacity and corner pip the chance, its width the gate length, the divisions the ratchets, a bridge the tie, a line the lane. Every row has its own playhead; steps that will not play this pass because of their chance are struck through. With LIVE the signal runs under the cells (input grey, output green — where they differ, the sequencer is acting), every row carries a thin trace of its measured gate (attacks, releases and levels as they really were), and a gate meter sits next to the row name.

Presets: a factory pattern library (Rhythm, Glitch, Space — Trance Gate, Stutter Roll, Tape Stop End, Dub Throws, Polymeter 5:7:3 and more) and your own .sfxp files in the seqfx_presets folder of the Omni config directory, with import and export. A preset stores the whole pattern together with the device parameters. Projects and presets made with the earlier Sequenced FX open and sound the same.

17. Utility

The everyday mixing tool: gain staging, stereo image and channel fixes in one slot, with a live goniometer. A neutral Utility passes the audio bit for bit, and every change glides click-free.

• Gain (−∞ to +24 dB; the bottom of the knob is silence) and MUTE (a click-free fade, the chain keeps running).
• Stereo: Width (0–400 %, the side level: 0 % is mono, 100 % untouched) and Balance (the centre is unity, the far side fades out, −3 dB half way).
• Channels (header keys): ST (stereo), SWAP (left and right exchanged), L or R (one channel on both sides) and MONO ((L + R) / 2).
• Phase: Ø L and Ø R flip the polarity of each output channel (also by clicking an output node of the routing display).
• Bass Mono: removes the side below the split (20–500 Hz) with a 24 dB/oct Linkwitz-Riley crossover; the mid goes through the matching all-pass, so the image above the split is exactly as before. Drag the handle in the BASS MONO strip to move the split, double-click to switch it on or off.
• Align: Offset delays one channel by up to ±20 ms (positive = right later), to line up two microphones or for a Haas effect; zero is the input bit for bit.
• DC removes DC offset below 5 Hz. LISTEN M / S plays only the mid (what mono keeps) or only the side (what mono loses).

The goniometer shows the output as a phosphor trail (mid up, left and right on the diagonals, auto-scaled) over the ghost of the transform: a reference stereo image passed through the engine's own matrix, so width, balance, routing and polarity are visible even without audio (red when the setting turns a mono signal out of phase); drag sideways for Width, double-click for 100 %. Under it the correlation bar (+1 mono, 0 wide, below 0 out of phase) with its recent low. The routing display draws the whole transform as wires from the inputs to the outputs (thickness = how much of that input arrives, red = inverted), the output level on each node, the offset as a delay tag and chips for the engaged stages. The meter bridge shows the input, the left and right outputs with peak hold, and W: the side over the mid in dB (red above 0 dB, where the image is close to out of phase).

Presets: a factory library (Essentials, Stereo, Low End, Repair, Gain) and your own .utilomni files in the utility_presets folder of the Omni config directory, with import, export and A/B compare.

18. Stereo Imager

Multiband stereo width: the image split into up to four bands, each with its own width, plus a mono-safe Stereoize, rotation of the whole image and mono / side listening. Both halves of the image (mid and side) go through identical Linkwitz-Riley filters, so at any setting the bands add back up flat; with one band at neutral settings the audio passes bit for bit.

• Bands (header keys 1–4) and the crossover slope (12, 24 or 48 dB/oct). Changing either lets the new filters settle in the background and crossfades to them, so it never clicks.
• Width per band (0–300 %, the side level: 0 % is mono, 100 % untouched). Typical: the lowest band at 0 % for a centred low end, the top bands above 100 % for air.
• Stereo (Stereoize) per band: the band's mid, delayed by ST Time (1–30 ms, shorter on higher bands so the combs never line up), is added to its side only. A mono source becomes wide, and in mono the effect cancels completely, unlike a plain Haas delay.
• Rotate turns the whole image (±45°), to fix a lopsided mix or lean it on purpose. Output ±24 dB.
• LISTEN MONO plays the mono fold-down, LISTEN S only the side. The S key on each band solos it.

The width field shows the bands over frequency: each band's width as a coloured level (lifted above the 100 % line or sunk below it), Stereoize as a dotted shimmer, the crossovers as lines with their frequency. Drag a band up or down for its width (Alt: all bands together, Shift: fine), turn the wheel over it for Stereoize, double-click to reset it; drag a crossover line sideways. The lane underneath draws the engine's own band responses (the slope is visible) and, with LIVE, each band's phase correlation; the output spectrum runs behind the bands. The goniometer draws one ghost per band (its width, the rotation and the listen mode) under the live trail; drag sideways for Rotate, double-click for 0°. Under it the correlation bar, on the right the meter bridge (input, left / right output, W = side over mid).

Presets: a factory library (Essentials, Mastering, Instruments, Creative, Repair) and your own .imgomni files in the imager_presets folder of the Omni config directory, with import, export and A/B compare.

19. Analyzer

A meter you can put anywhere in a chain: after one instrument, on a group, before and after the master limiter. Audio passes bit for bit, it never changes the sound. It measures loudness to ITU-R BS.1770-4 / EBU R128 and true peak, and draws the spectrum and the goniometer of exactly what leaves its slot.

• Target (header keys): STREAM -14 LUFS / -1 dBTP (Spotify, YouTube, Tidal, Amazon), APPLE -16 / -1, EBU -23 / -1, ATSC -24 / -2, CLUB -9 / -0.3, CUSTOM (the Target and Ceiling knobs). Readings are coloured against it: green within 1 LU, red louder (the platform will turn it down), amber quieter.
• Loudness: integrated (gated over the whole measurement, shown large with its distance to the target), short-term (3 s), momentary (400 ms) with their maxima, loudness range (LRA, with its 10th and 95th percentile), true peak maximum per channel (4x oversampled, so inter-sample peaks count), sample peak, PLR (true peak minus integrated) and the measured time. The history graph under the numbers draws short-term and momentary loudness over 10 s to 10 min with the target band and the integrated line.
• Meters: momentary and short-term bars around the target, true-peak bars of both channels with the ceiling line and the held maximum.
• RESET starts a new measurement; with AUTO on, every play start does it, so each playthrough is measured on its own. Stopping or jumping on the timeline never clears it.
• Spectrum: resolution 2K to 32K points (the FFT runs on the interface thread, it costs the audio engine nothing), Response (averaging time), Slope (tilt in dB/oct around 1 kHz; at 4.5 a well-balanced mix reads roughly flat), Top and Range of the scale, VIEW mix, mid / side or left / right overlaid, HOLD off, 1 s, 3 s or infinite, FREEZE to hold the picture still. Drag up or down for Top, wheel for Range, double-click clears the peak holds; the crosshair reads frequency, note and level.
• Goniometer and correlation bar: the stereo image of the output, its phase correlation and side-to-mid ratio.

Presets: a factory library (Delivery, Mixing) and your own .anaomni files in the analyzer_presets folder of the Omni config directory, with import, export and A/B compare. Loading a preset never restarts the measurement.

20. Dynamic EQ

Six EQ bands that move with the music: each one cuts or boosts only while its part of the spectrum is too loud or too quiet. The bands are the same matched bells and shelves as the EQ (no cramping near the top of the spectrum); with every band off the audio passes bit for bit.

• Band: Shape (BELL, LO shelf, HI shelf, TILT), Freq, Gain (the static part), Q.
• Dynamics: ABOVE acts while the band's level is over the Threshold, BELOW while it is under it, OFF keeps the band static. Range is the furthest the band moves and its sign is the direction: negative cuts, positive boosts. So one band can tame a resonance (Above, minus), lift presence in quiet passages (Below, plus) or let air through only when it is loud (Above, plus). Ratio sets how fast the range is reached, Attack and Release how quickly the band moves; a 6 dB soft knee around the threshold.
• Routing: each band works on ST (both channels), M, S, L or R, and keys from its own region of the input (SELF) or of the external sidechain (EXT, the source is picked in the header). LISTEN plays what the band's detector hears.
• Global: DETECT peak or RMS, Depth (scales every range at once, 0-200 %), Lookahead (0-20 ms, reported to the delay compensation), Mix, Output.

The graph shows each band's static curve, the region its dynamics can reach and, with LIVE, the curve it plays now, plus the summed curve and the input / output spectra. Drag a node for frequency and gain (Shift: fine), drag its diamond for Range, turn the wheel over a node for Q, double-click empty space for a new band and a node to switch it off. The gain computer on the right draws the selected band's transfer curve with the live level riding on it: drag sideways for Threshold, up or down for Range. The band tabs select a band; their LED switches it on or off and a bar shows how far it moves.

Presets: a factory library (Mixing, Vocals, Instruments, Mastering, Sidechain) and your own .deqomni files in the dynamic_eq_presets folder of the Omni config directory, with import, export and A/B compare.

21. De-esser

Turns down sibilance (s, sh, t) and other sharp high-frequency bursts only at the moment they happen.

• Key: the detector listens to everything above the Frequency (HP) or to a band around it (BP, width Q). The key display shows that region in amber over the input / output spectrum; drag sideways for the frequency, wheel for Q, double-click to switch HP / BP.
• Detection: ABS compares the key band's own level with the Threshold; REL compares it with the whole signal, so one threshold keeps working when the singer moves closer or further away (start around -14 dB). Range is the deepest reduction, Ratio how hard it acts, Attack / Release how fast; a 3 dB soft knee.
• SPLIT turns down only the sibilant region through a dynamic high shelf (HP key) or bell (BP key), the EQ's own filters; with no sibilance the audio is untouched. WIDE turns the whole signal down on each ess, the classic broadcast sound.
• Channel ST, M (the centre of a mix, where the lead vocal sits) or S (wide fizz, reverb splash). The key can come from the external sidechain (picked in the header), for example a dry vocal while the de-esser sits on a vocal bus.
• Lookahead (0-10 ms, reported to the delay compensation) lets the reduction start before the ess. LISTEN KEY plays what the detector hears, LISTEN DELTA only what is removed: if you hear words in it, the setting is too strong.

The history on the right scrolls the detection level against the dashed threshold line (drag it up or down, double-click for -30 dB) with the reduction hanging from the top in red and a reduction meter with hold. The red curve in the key display is the filter at the reduction playing now, its full range as a faint ghost.

Presets: a factory library (Vocals, Instruments, Mix & Master) and your own .essomni files in the deesser_presets folder of the Omni config directory, with import, export and A/B compare.

22. Flanger

A short delay swept under the dry signal: the comb of notches it carves moves through the spectrum for the classic jet and whoosh.

• CLASSIC: the delay sweeps from Delay up to Delay + 8 ms at full Depth.
• THRU-0 (through-zero): the dry signal is delayed by 10 ms and the wet sweeps across it, so the two meet at zero delay: the deep "tape jet" null of two tape machines. The 10 ms are reported to the delay compensation, so the track stays in time. Delay sets the swing here.
• BARBER (barberpole): two voices ramp in one direction and fade in and out, so the notches rise (UP) or fall (DOWN) for ever without turning around.
• Rate free in Hz, or SYNC to a note value locked to the song position (the same sweep in every bounce). SINE or TRI LFO. Stereo Phase offsets the right channel (180° = opposite sweeps).
• Feedback ±95 % (positive: ringing peaks, negative: hollow notches) with Damping in the loop, Mix, Output. At Mix 0 the audio passes bit for bit.

The comb display draws the frequency response playing now for left and right with the notches marked: drag sideways for Delay, up or down for Feedback, double-click for no feedback. The sweep view draws the delay over two LFO cycles (barberpole voices fading, the through-zero dry line) with the playhead: drag up or down for Depth, wheel for Rate, double-click for sine / triangle. Without LIVE both run on a preview clock.

Presets: a factory library (Classic, Through-Zero, Barberpole, Synced) and your own .flngomni files in the flanger_presets folder of the Omni config directory, with import, export and A/B compare.

23. Clipper

A waveshaping peak limiter: it cuts the peaks off at the ceiling instantly, with no attack or release, so drums keep their punch where a limiter would pump. Typical uses: shaving transients before the master limiter, cracking a snare, saturating a bass.

• Style: HARD (flat at the ceiling; Knee rounds the corner from a sharp edge to a smooth bend), SOFT (a rounded cubic curve, mostly odd harmonics) or ANALOG (an asymmetric tube-like curve with even harmonics, with a DC blocker).
• Input drives the signal into the curve, Ceiling (-24 to 0 dB) is the level it never exceeds inside the clipper. AUTO gain undoes the input gain at the output, so you hear the clipping rather than the extra level.
• OS oversampling 1x (a raw digital clip, aliasing on purpose), 2x, 4x or 8x (the cleanest). The latency it adds is a whole number of samples, reported to the delay compensation; the clean signal is delayed by exactly the same amount, so Mix (parallel clipping) and DELTA (hear only what is clipped away) line up sample for sample. Changing the factor fades out and back in.
• ST clips left and right, M/S the mid and side separately.

The transfer display draws the curve at the current input and ceiling against the straight line and the ceiling lines, with the live input peak on it and the clipped part in red: drag up or down for Ceiling, sideways for Input, double-click for 0 dB input. The clip history scrolls the driven input (grey) and the output (green) against the ceiling with everything pushed over it in red, and a reduction meter with hold and the share of clipped samples on the right.

Presets: a factory library (Mastering, Drums, Color) and your own .clipomni files in the clipper_presets folder of the Omni config directory, with import, export and A/B compare.

24. FX Rack

A container for parallel processing: up to six branches, each its own chain of built-in effects, mixed back into one signal inside a single slot. Typical uses: parallel compression, effects on one frequency band only, reverb on the sides only, a different echo per side.

• Split decides what each branch hears: PAR (every branch gets the whole signal, the branches are summed), BAND (branch by branch from low to high, split by Linkwitz-Riley crossovers that sum flat; slope 12 / 24 / 48 dB/oct, drag the frequency pills between the lanes, Shift for fine), L/R (the first branch gets the left channel, the second the right, each as mono) or M/S (the first branch gets the mid, the second the side).
• Every branch has a name (double-click), Level, Pan, Mute, Solo and Ø polarity, a meter and its latency. + opens a new branch, × closes one with its devices.
• Devices: + Add puts any built-in effect into the branch (except Convolution Reverb, Neural Amp, Sequenced FX and plug-ins, which stay on the track's own chain); up to 16 per rack. Click a device to open its own panel next to the rack, the power dot bypasses it, drag it to move it within or between branches, right-click for Earlier / Later, Move to branch, Duplicate and Remove.
• Latency is lined up across branches: a branch with a linear-phase EQ or an oversampled clipper does not smear the sum, and the rack reports the slowest branch to the delay compensation.
• Macros: eight knobs that each drive any number of parameters. Press MAP on a macro, then touch any control in the rack or in the open device: it is mapped over its whole range. The map editor shows every map of the macro with its own from and to values (frequencies and times glide on a log scale), ⇄ to turn it the other way and × to remove it, plus the macro's name. DONE leaves MAP mode.
• Mix blends the rack with the dry signal (delayed by the same latency), Output sets the level.

Every parameter inside the rack, the devices' included, can be automated, MIDI-learned and modulated like any other: they appear in the automation list as branch · device · parameter. Editing the rack's structure keeps the devices that stay running, tails included.

Presets: a factory library (Parallel, Bands, Stereo) and your own .rackomni files in the fx_rack_presets folder of the Omni config directory, with import, export and A/B compare. A rack preset stores the whole rack: branches, devices, macros and their maps.

25. Tremolo

An LFO that moves the level, the stereo position, or the low end against the top. Typical uses: amp tremolo on guitars and keys, rhythmic gating locked to the song, a sidechain-style pump without a sidechain, auto-pan.

• Mode: TREM (the level of both channels), PAN (the balance, equal power: the centre is untouched, a full swing reaches −∞ on one side and +3 dB on the other) or HARM (harmonic tremolo: lows and highs pulse in turn across a 24 dB/oct crossover, set with Crossover).
• Shape: SIN, TRI, SQR (on and off), UP (rising saw, the pump), DN (falling saw) and RND (a new level every cycle, repeating with the song). Symmetry moves the dip earlier or later (the square's duty), Smooth rounds the square's edges, the saws' resets and the random steps.
• Rate in Hz, or SYNC to a note value locked to the song position; Phase shifts the cycle against the bar. Stereo offsets the right channel's cycle (180° = left and right in turn).
• Style: CLEAN follows the LFO exactly; OPTO adds a photocell's lag, the dip lands fast and recovers slowly, as in the classic optical amp tremolo.

The cycle view draws two LFO cycles exactly as the mode plays them (left and right, low and high, or the balance) with the depth band and the playhead: drag up or down for Depth, sideways for Symmetry, the wheel for Rate, double-click for the next shape. The motion view scrolls the last two seconds: the level of each side (or band) as an envelope, or in Auto-pan the stereo image with its trail. Synced, the displays follow the song even without LIVE. At zero depth Tremolo and Auto-pan pass the signal bit for bit.

Presets: a factory library (Classic, Rhythmic, Auto-pan) and your own .tremomni files in the tremolo_presets folder of the Omni config directory, with import, export and A/B compare.

26. Frequency Shifter

Moves frequencies instead of pitch, or multiplies the signal by a carrier. Typical uses: endless phasing and barber-pole motion, a chorus without a delay, metallic and bell-like timbres from ordinary sounds, robot voices, climbing or falling feedback spirals.

• Mode: SHIFT moves every frequency by the same number of hertz (single sideband through an analytic signal, the opposite sideband sits near −90 dB). Harmonics stop lining up, so a few hertz give phasing and a few hundred turn a guitar into a bell. RING multiplies by the carrier: sum and difference tones, the original gone. AM keeps the original under both sidebands.
• Direction: UP or DN (Shift). Shifting down folds the partials below the carrier through zero.
• Carrier shape (Ring and AM): SIN, TRI, SQR (odd sidebands) and SAW (every sideband), with band-limited edges.
• WIDE runs the right channel's carrier the opposite way: in Shift, up on the left and down on the right. Phase offsets the right channel's carrier.
• LFO: Rate in Hz, or locked to a note value of the song; Depth in percent of the frequency. Past 100 % the carrier swings through zero.
• Feedback: Amount sends the wet signal back in after Delay (0 ms = the shortest loop, one sample), through the Tone low-pass. Every trip shifts again, so a tail climbs or falls further on each echo. A DC blocker and a gentle limit keep the loop in check. A new delay crossfades without a pitch bend.
• GUARD band-limits the input so nothing is pushed past Nyquist (no aliasing). Its cut-off follows the carrier.
• Mix (0 = the signal passes bit for bit) and Output.

The map shows where the frequencies go. A G3 and its harmonics stand on top; below, every component heard, with a thread from where it came. A ring marks a partial folded through zero; red marks one aliased past Nyquist. Drag sideways for Frequency, use the wheel to fine-tune, double-click to flip the direction (Shift) or change the carrier shape (Ring / AM). The spiral shows the feedback loop trip by trip: each echo one Delay later, moved again and quieter by the loop's gain at its new frequency. Drag up or down for Feedback, sideways for Delay. With LIVE the map reads the engine's carrier as it moves.

Presets: a factory library (Shift, Ring & AM, Spiral) and your own .shiftomni files in the freq_shifter_presets folder of the Omni config directory, with import, export and A/B compare.

27. Bitcrusher

Lowers the sample rate and the bit depth. Typical uses: vintage sampler grit, telephone and radio sound, chip-tune crunch, lo-fi beats, digital destruction mixed in under a clean signal.

• Rate from 48 kHz down to 100 Hz. The hold's steps land at the exact instants of the new rate (band-limited), so the only aliasing you hear is the one the new rate makes. Jitter stretches and shrinks every period (the same way on every pass). Stereo runs the right channel's clock up to a quarter octave faster, so the aliases move apart.
• Bits from 24 down to 1, fractional values included, so the depth glides. LIN gives even steps; MU-LAW gives fine steps near silence and coarse ones at the top, as in telephone codecs; FLOAT keeps a short mantissa, so the step follows the level. Dither adds triangular noise before the rounding.
• Overflow: past full scale the signal is CLIPped, WRAPs round to the other side like integer overflow, or FOLDs back. Drive hits the quantizer and the overflow harder; the level is given back afterwards.
• Alias: 100 % lets every fold through, 0 % filters the input at the new rate first. Smooth: 0 % keeps the stairs, 100 % filters the images off.
• Mix (0 = the signal passes bit for bit, one sample late: the hold's latency, compensated by PDC) and Output.

The wave view runs a sine through the engine with your settings: the stairs of the new rate, the quantizer's levels, jitter, folds and smoothing. Drag up or down for Bits, sideways for Rate, use the wheel to fine-tune the bits, double-click to change the law. The spectrum shows where an A3 and its harmonics land: kept (blue), folded below half the new rate (red) and images round every multiple of the rate (amber). Drag sideways for Rate, up or down for Alias.

Presets: a factory library (Vintage Digital, Lo-Fi, Destroy) and your own .crushomni files in the bitcrusher_presets folder of the Omni config directory, with import, export and A/B compare.

28. Tape

A tape machine: saturation that remembers where the signal came from, the low-end bump of the playback head, the softened top, wow and flutter, hiss and the dropouts of an old reel. Typical uses: glue on the mix bus, weight on drums and bass, warmth on vocals, the wobble of a cassette.

• Speed: 3.75, 7.5, 15 or 30 ips. The speed sets the head bump frequency, how early the top falls, the record emphasis, how fast and how much the tape wavers, and the hiss.
• Drive pushes the tape: the signal runs through a magnetic hysteresis loop at twice the sample rate, with the highs lifted before it and lowered after, so they saturate first, as on a real machine. Bias: less brightens and dirties the tape, more cleans it and dulls the top. AUTO gives the level back after Drive.
• Bump: the head bump of the speed, with the small dip above it and the low end falling away below.
• Wow (slow) and Flutter (fast) bend the pitch; a slow tape wavers more. Stereo pulls the channels apart, so the image moves.
• Hiss, and Wear: dropouts that dip and muffle the sound for a moment, and a duller top.
• Mix (0 = the signal passes bit for bit, later by the latency) and Output. The latency (the oversampling plus 4 ms between the heads) is compensated by PDC.

The loop view draws a −3 dBFS sine through the tape: input across, output up. It opens and bends as Drive rises and Bias falls; drag up or down for Drive, sideways for Bias. The response shows the playback curve: drag up or down for Bump, double-click for the next speed. The pitch strip below it scrolls the last three seconds of wow and flutter for both channels: drag up or down for Wow, sideways for Flutter.

Presets: a factory library (Studio, Colour, Worn) and your own .tapeomni files in the tape_presets folder of the Omni config directory, with import, export and A/B compare.

29. Spectral

Takes the sound apart into hundreds of frequency bins and works on each one: a spectral gate, blur in time, freeze and smear across frequency. Typical uses: a drone or pad held from any moment, a wash that turns a part into a cloud, hazy or breathy textures, cleaning noise and rooms out between the notes, or the opposite: keeping only the air.

• Size: 1K, 2K or 4K points. Small frames react fast but sound grainy and robotic; large ones are fine in frequency and smooth in time.
• Gate: each bin is compared with the Threshold (both channels read together, so the image stays put). BELOW cuts the quiet bins (noise, room, hiss), ABOVE cuts the loud ones (the tones go, the air stays). Range sets how deep, Release how fast a bin closes, Tilt slopes the threshold around 1 kHz (negative = lower threshold for the highs, where noise is quieter).
• Blur smooths every bin in time, from 30 ms to 8 s: transients melt and every note rings on at its own frequency.
• Freeze holds the spectrum and lets every bin ring on at its own pitch; Fade sets the crossfade in and out, Drift lets the held bins wander by up to 30 cents, like a choir. The held spectrum survives stopping the transport; the panic (Stop twice) clears it.
• Smear spreads every bin over up to two octaves at the same loudness: tones turn into bands of noise, voices into breath.
• Low and High keep all of this to a band; the rest passes untouched.
• Mix (0 = the signal passes bit for bit, later by the latency) and Output. The latency (the frame size) is compensated by PDC.

The spectrum shows the engine's frame per third octave: the input filled with its peaks held, the output as the bright line, the held spectrum in ice, and what the gate takes hanging from the top, under the dashed threshold. Drag up or down for Threshold, drag an edge sideways for Low or High, double-click to Freeze. The spectrogram beside it scrolls the output: blur smears to the right, smear up and down, the gate leaves holes. The lens below shows one click of one tone through Blur (across) and Smear (up and down); drag to change them.

Presets: a factory library (Freeze, Texture, Gate) and your own .specomni files in the spectral_presets folder of the Omni config directory, with import, export and A/B compare.

30. Pitch Shifter

Up to four voices from one input: shifted copies for octaves and doubles, or a harmonizer that sings thirds and fifths in key with whatever note it hears. Typical uses: an octave under a bass or a lead, a double track, backing harmonies from one vocal, a smaller or bigger voice, special effects.

• Voices A–D (the keys in the header): each has Interval, Detune (cents), Level and Pan.
• CHROMATIC: the interval is in semitones (±24): every note moves by the same amount.
• SCALE: the interval counts degrees of the Key and Scale from the note the device hears, so "+2" is a major or a minor third as the key wants. When no note is heard (a breath, a consonant) the last interval holds. Glide smooths the moves between intervals.
• KEEP formants: the voice stays itself at any pitch (no chipmunk). Without it the formants ride with the pitch, like tape sped up or slowed down. Formant (±12 semitones) moves them on its own: a smaller or a bigger throat at the same pitch.
• Size: 1K, 2K or 4K points: small frames suit drums and fast parts, large ones low voices and pads.
• Mix (0 = the signal passes bit for bit, later by the latency) and Output. The latency (the frame size) is compensated by PDC.

The harmony roll shows the last seconds on the rows of the key: the note heard in white and where every voice takes it in its own colour; on the right, the dots of "now". Drag a dot up or down to change that voice's interval. The formant view shows the input's envelope and the one the first voice comes out with; drag sideways for Formant, double-click for Keep. Without LIVE it shows a model vowel.

Presets: a factory library (Shift, Harmony, Character) and your own .pitchomni files in the pitch_presets folder of the Omni config directory, with import, export and A/B compare.

31. Pitch Correction

Tunes a voice (or any monophonic part) to the notes of a key: from a gentle touch nobody hears to the hard, stepped effect. Typical uses: fixing a few off notes, tightening backing vocals, the robotic pop sound, tuned octave or gender effects.

• Key and Scale set the notes a pitch may land on; the note strip (C to B) switches single notes off, e.g. to keep a melody away from a note that clashes. With every note switched off, all notes are allowed. The default scale is Chromatic: every semitone is a target.
• Speed: how fast a note is pulled to its target (0 = at once, the hard effect). Humanize slows the pull the longer a note is held, so held notes keep their life while quick ones still land. Amount sets how much of the way it goes, Tolerance leaves notes that are already that close alone.
• Transpose (±12 semitones) and a synthetic Vibrato (depth in cents, Rate) ride on top.
• KEEP formants: the voice stays itself when moved; Formant (±12 semitones) changes the throat on its own.
• Size: 1K (least latency, high voices), 2K or 4K (smoothest on low voices).
• Mix (0 = the signal passes bit for bit, later by the latency) and Output. The latency is compensated by PDC.

The pitch graph shows the last seconds on the rows of the allowed notes: the pitch heard (pale), what comes out (bright) and the target note (a bar). Drag sideways for Speed, up or down for Amount. The tuner beside it shows the target note, how many cents off the pitch heard is, the correction, and the response: a note sung 45 cents sharp and how these settings bring it home. Click a key of the strip to allow or leave out that note.

Presets: a factory library (Natural, Hard, Creative) and your own .tuneomni files in the tune_presets folder of the Omni config directory, with import, export and A/B compare.

32. Vocoder

The classic talking synth: the voice on the track shapes a carrier band by band, so the carrier speaks with the voice's words. Typical uses: robot and synth voices, choirs from one vocal, a pad or a synth playing the rhythm of a drum loop, whispered textures.

• Carrier: INT plays the built-in synth, SC takes another track through the sidechain picker on the right (while none is routed, the built-in synth plays).
• The built-in synth: SAW, SQR, SUPER (three detuned saws, wide) or NOISE (a whisper). FOLLOW plays the note the voice sings (+ Transpose), FIXED stays on Note. Noise adds noise to any carrier for clearer consonants.
• The bank: Bands (8 to 40: few = coarse and synthetic, many = clear words) between Low and High, Width of each band, Formant (±12 semitones) moves the carrier's bands: a smaller or a bigger "mouth".
• Attack and Release set how fast the bands follow the voice, Emphasis tilts them (more highs = more presence). Unvoiced mixes the voice's own top back in so "s" and "t" stay clear. Freeze holds the voice's last shape.
• Every carrier band is normalised to its own level, so the result has the voice's spectrum and level whatever the carrier.
• Mix (0 = the signal passes bit for bit) and Output. No latency.

The bands view shows, per band, the voice's level (up) and the carrier band that plays it (down), joined by threads that lean with Formant; drag sideways for Formant, up or down for Width, double-click to Freeze. The bank view draws the filter bank's response with the Low and High edges (drag them) and what the carrier is right now.

Presets: a factory library (Classic, Robot, Texture) and your own .vocomni files in the vocoder_presets folder of the Omni config directory, with import, export and A/B compare.

33. AI Denoiser

Removes background noise (fans, hiss, hum, room noise, wind) and leaves the voice or instrument. Two engines:

• AI: a recurrent neural network (RNNoise) tells voice from noise, best on speech. Amount (0 to 60 dB) is how much noise is taken off; Voice Gate pulls everything down between words when the network hears no voice.
• SPECTRAL: subtracts a noise profile bin by bin, works on any material. PRINT uses a learned noise print: press LEARN over a second or two of the noise alone, then stop (the print is saved with the project; the bin icon clears it). ADAPT tracks the noise floor from the signal as it changes, no print needed (very long held notes slowly count as noise; use a print for them). Threshold moves the noise line, Smooth trades speed for fewer artefacts, 1K / 2K / 4K set the frame size (fast transients or fine hum lines).
• Mix (0 = the signal passes bit for bit, delayed by the device's latency like the wet signal) and Output. LISTEN plays only what is taken away, to check nothing of the voice leaks.

The spectrum shows the input, the output, what is removed (red), the noise line and the print; drag up or down for Threshold (Spectral) or Amount (AI), double-click to Learn. The side view shows the suppression law with every band as a dot (Spectral; drag for Amount and Threshold) or the network's voice probability with the gate line (AI; drag for Voice Gate).

Presets: a factory library (Voice, Noise Print, Music) and your own .denomni files in the denoiser_presets folder of the Omni config directory, with import, export and A/B compare. Presets keep your noise print.

34. Neural Amp (NAM)

Plays Neural Amp Modeler captures (.nam files: amps, pedals, full rigs) with a cabinet impulse response in the same device. Click the amp plate or drop a .nam on it; click the cabinet view or drop an audio IR (WAV, AIFF, FLAC, OGG, MP3; mono, stereo or true stereo) on it.

• Gate (−80 dB = off) runs before the amp, Input drives it. MONO runs one model on the summed input (half the CPU), STEREO one per channel.
• AUTO LEVEL brings every capture to −18 dBFS: from the capture's own loudness data, or measured when the file has none. Switching models no longer jumps in level.
• Bass / Mid / Treble shape the tone after the amp, then the cabinet (zero latency, up to 0.5 s, normalised so cabinets swap at the same level; the trash icon removes it, CABINET switches it), then Low Cut and High Cut.
• Output and Mix (0 = the dry signal). The device adds no latency.

The amp plate shows the model, Auto Level's trim and the levels into the amp, out of it and at the output; it warns when the capture was made at another sample rate than the engine runs at (the tone shifts). The cabinet view draws the cabinet's response with the cuts; drag an edge sideways for Low Cut / High Cut. The cabinet's file is saved with the project.