Ghost Dynamics

The level boundary shared by downward and upward behavior. Lowering it makes more incoming energy eligible for Down; raising it changes the region where Up can recover quieter information. Source and detector filters determine what is measured.

Try this: With Down 50% and Up zero, lower Threshold until the loud events are visibly controlled.

Scales the overall dynamic transformation without changing the relative Down/Up balance. Zero removes dynamic gain; 100% applies the chosen relationship fully. It is not simply a ratio or an output-volume control.

Try this: Find a useful Down/Up setup, then reduce Depth to back off the whole transformation.

Controls above-threshold restraint. Zero is inactive; increasing it moves Clean from ordinary compression toward strong peak control. A low enough Threshold and an active detector are necessary for a noticeable result.

Try this: Start around 50% on a dynamic loop before trying extremes.

Recovers coherent quieter detail below Threshold with bounded, salience-aware behavior. Zero is inactive. It is not a semantic noise-removal classifier: unusual background material can still be brought forward. Gate closure takes priority over competing Up.

Try this: Add 10-20% after setting Down, comparing quieter phrases and background level.

The main response timescale. Short values follow changes quickly; long values retain energy and recover more slowly. It still influences deeper detector weighting and Character memory when Precision Attack/Release are active; it does not control the separate gate timing.

Try this: Compare 40 and 200 ms on a repeated phrase with Down already active.

Moves between Transient on the negative side, Balanced at zero and Body on the positive side. It changes the importance of leading edges versus sustained material in the response, not a separate transient-effect chain.

Try this: Compare -25% and +25% on drums while keeping Time fixed.

Moves continuously from unified broadband dynamics to three internal frequency regions. Higher values permit frequency-dependent response without user crossover, per-band threshold or per-band attack controls. At zero, begin with familiar broadband behavior.

Try this: Raise to 50% after establishing a useful broadband setup.

Controls cohesion between the three regions. Higher values draw their responses together; lower values allow greater independence. Its role becomes more evident as Bands increases.

Try this: At Bands 75%, compare 100% link and 30% on bass-heavy material.

At 100%, uses a fully linked response for image stability. At zero, left/right dynamics can genuinely differ for both Down and Up; intermediate values interpolate. It does not synthesize width or pan the source.

Try this: Start at 100%. Reduce only when program-dependent image movement is intended; mono cannot create stereo behavior.

Main derives control from the program. Sidechain uses the external detector input without mixing it into the audible signal. A missing external bus is explicitly unavailable/silent rather than silently replaced by Main.

Try this: Use Main for the first compression test. Choose Sidechain only after routing a real detector source in the DAW.

Chooses Clean, Punch, Glue, Elastic, Density or Breath. These change dynamics/envelope behavior and supporting activity-dependent coloration, not six arbitrary saturation effects. Changing modes preserves populated history through a safe handoff.

Try this: Compare Clean and Punch on drums with Down already active; modes need a dynamic transformation to reveal their role.

Blends latency-aligned dry and processed signal. Zero and 100% are exact endpoints. Equal-power intermediate blends can raise level when correlated, so judge the envelope after level comparison.

Try this: Try 50% for parallel compression after identifying the fully wet result.

Final trim after Mix. It does not change detection sensitivity, Threshold or gain-reduction strength, and is not automatic makeup gain.

Try this: Raise it modestly if compression lowers total level, then compare bypass at comfortable loudness.

Enables manual global Attack and Release. Off uses the main Time/Contour relationship. Saved Attack/Release values remain while it is off; gate timing is independent.

Try this: Leave Off for the first setup, then enable it only when explicit opening/recovery timing is needed.

Manual global gain-response attack when Precision is on. Short values react faster; longer values permit more leading edge. Time still shapes the broader memory behavior.

Try this: At Precision On compare 3 ms and 25 ms on a drum loop.

Manual global recovery when Precision is on. Short settings recover sooner; longer settings join events into a slower response. It is separate from Gate Release.

Try this: Compare 100 and 300 ms with the phrase's gaps.

Introduces ordinary below-threshold downward expansion, progressing toward strong closure. Zero disables this action. It shares Threshold/Source/linking and suppresses competing Up while closed.

Try this: Begin at 25% with a source containing real gaps, then increase if more closure is needed.

Caps how much the gate can attenuate. A small range gives gentle reduction; a large range permits deep closure. It matters only when Gate Amount and detector conditions call for attenuation.

Try this: Try 12 dB before using the full default range.

Controls how quickly the gate opens after an event. Very short values preserve an attack; longer values soften the opening. It does not use the Precision switch.

Try this: Start near 1 ms for percussion and increase only if a softer entrance is desired.

Keeps the gate open briefly after an event before closing can proceed. Longer hold can join nearby hits and avoid cutting between them.

Try this: Try 50-100 ms on a choppy part before changing Gate Release.

Controls the gate's closing movement after the hold. Short values cut gaps decisively; longer values preserve more decay. This is not global compressor Release.

Try this: Try 150 ms to retain some room or instrument decay.

Separates opening and closing boundaries so small detector fluctuations do not repeatedly toggle the gate. Higher values discourage chatter but change when closure occurs.

Try this: Start with 3 dB, then increase modestly if the gate chatters near Threshold.

High-pass filtering for both Main and Sidechain detection, not audible program EQ. Raising it reduces low-frequency influence on the response; default 20 Hz is off-equivalent.

Try this: Try 100 Hz if bass dominates detection, then compare with 20 Hz.

Low-pass filtering for both detector sources. Lowering it reduces high-frequency influence without directly darkening program audio; the resulting dynamics may still change its balance.

Try this: Try 6 kHz if bright transients dominate control.

Changes detector sensitivity before dynamics, not audible output gain. Higher values make the detector see more energy; lower values less. It interacts with Threshold and selected Source.

Try this: Leave at zero for normal setup; use modest trim to align an external detector's working level.

L/R operates on left/right information. M/S operates on existing center and side information; independent treatment can alter their balance without inventing Side content or acting as a panner. Stereo-only behavior is inactive in mono.

Try this: Begin L/R. Explore M/S only with stereo material and careful linking/level comparison.

Delays the audible path so detection can anticipate an event. Uses its actual sample-rounded delay, not a constant maximum. It changes host-reported latency and aligned dry/wet timing; it does not need to be on for normal compression.

Try this: Start at zero. Add 1-3 ms for anticipatory peak control when that trade-off is useful.