User Manual
Version 1.0
Bipolar compressor based on ButterComp2 by Airwindows.
Original source code published under open license by Chris Johnson.
Air-G Velvet is a bipolar compressor based on the ButterComp2 algorithm by Airwindows. Its operation is unlike any conventional compressor: rather than acting uniformly on the waveform, it treats the positive and negative halves separately, balancing the shape of the wave while compressing. Internally, four compressors work in parallel.
The result is a very gentle and musical compression: it adds density, glue, and a characteristic tonal reshaping. It softens high-frequency detail in a particular way, and has a remarkable effect on ambiences and reverb tails — making them float in space in a way that is difficult to achieve with other compressors.
In some respects it behaves similarly to classic Vari-Mu tube compressors (Fairchild 670, Manley Vari-Mu): it responds to the musical program without needing explicit attack or release settings, it is gentle and hard to misuse, and it adds glue and cohesion. But its internal mechanism — the bipolar treatment and four parallel compressors — is its own, with no equivalent in analog hardware.
It works well in almost any context: master bus, drum bus, vocals, guitars, bass, ambiences, reverbs.
Air-G Velvet interface.
The blue LED indicates that the plugin is receiving audio and processing. It turns off when the DAW bypasses the plugin or when there is no active signal. It is informational only — it has no interactive function.
The LED has two behaviors: when the output approaches the limit (above -0.8 dB), it starts to glow gradually. This glow is not persistent — it follows the signal in real time. If the output exceeds 0 dBFS, the LED latches at full brightness and stays on until the user clicks it.
This allows the proximity to the limit to be seen in real time, and intermittent clipping to be detected that would otherwise go unnoticed.
If the LED latches regularly, consider lowering OUTPUT or reducing COMPRESS.
The main control. The higher the COMPRESS setting, the more the plugin compresses. At 0% it does nothing; as you raise it, the sound becomes denser, more glued, rounder.
An important characteristic of Velvet: there is no fixed ratio. The amount of compression depends on both the COMPRESS value and the incoming signal. At soft levels compression is nearly imperceptible; at louder levels the compressor works harder. That is what makes it sound natural.
Start between 30% and 50%. The 30–70% range is where the plugin sounds best in most cases.
Controls the output volume. When compressing, the overall level tends to drop slightly because the louder peaks are being reduced. OUTPUT lets you recover that level or adjust the output to taste.
At 50% the output is at unity gain. Above 50% the plugin amplifies the output; below 50% it attenuates it.
Blends the original signal (Dry) with the compressed signal (Wet). At 100% only the compressed signal comes through, which is the standard behavior. As you lower DRY/WET, the unprocessed original signal is mixed in with the compressed one.
This technique is known as parallel compression and is one of the most widely used in professional mixing. It allows you to add density and punch without losing the original transients: the compressed signal provides the body, and the original signal provides the attack.
Push COMPRESS to maximum, bring DRY/WET down to 40–60%. The result is a drum track that retains the punch of the original but with more energy and glue on the medium and soft hits. A classic drum mixing technique.
Filters low frequencies from the compressor’s detector — the signal the compressor “listens to” when deciding how much to compress. It does not affect the audio heard at the output.
Its main purpose: when there is a strong kick or bass in the mix, they can cause the compressor to react to every hit and produce an unwanted pumping effect. Raising SC HPF prevents the compressor from reacting to those low frequencies.
On vocals and guitars: keep it at minimum. On drum bus or master bus with heavy low-frequency content: raise it to 60–80% to stabilize the compressor’s action.
Selects which part of the stereo signal gets compressed. The knob has three fixed positions.
Standard mode. The compressor acts independently on the left and right channels. For general use: buses, master bus, vocals, instruments.
Only the center of the stereo image is compressed: vocals, kick, bass — everything that sits in the middle. What sounds to the sides is left untouched.
Useful when you want to control the center of a mix without changing the stereo width.
Only the lateral information is compressed. The center is left untouched. Compressing the SIDE signal narrows the stereo image during moments of high lateral energy.
The plugin handles all M/S encoding and decoding internally. No external encoder or decoder is needed. When changing modes, the compressor state resets automatically to prevent clicks or artifacts.
When active, the plugin automatically adjusts the output volume to compensate for the level reduction caused by compression. The result is that the output level stays approximately constant regardless of how much COMPRESS is raised.
Off (default): classic behavior. Raising COMPRESS lowers the level; OUTPUT must be adjusted manually to compensate.
On: level stays stable. Very useful for quickly comparing different COMPRESS settings without the volume changing.
For a more accurate evaluation of how much the plugin is compressing, keep AUTO GAIN off and watch the VU meters. With AUTO GAIN on it is easier to hear the tonal and dynamic effect of the compression without the volume difference interfering with the evaluation.
The two meters show how much the compressor is reducing the volume at any given moment. When there is no compression, the needle sits at 0. When the compressor acts, the needle moves right to show the dB of reduction.
Which meter is active depends on the selected mode:
COMPRESS 30–50%, OUTPUT adjusted to match the unprocessed volume, DRY/WET 100%. The typical result is a more cohesive mix: instruments sound together without any one disappearing. SC HPF at 80–120 Hz if there is heavy low-frequency content.
COMPRESS 50–70%, DRY/WET 50–80%. Parallel compression works very well on drums: it adds energy to medium and soft hits without crushing everything. SC HPF at 60–100 Hz.
COMPRESS 25–50%, DRY/WET 100%, SC HPF at minimum. The compressor controls peaks without sounding pumped. For layered vocals or stacks: DRY/WET 60–80% for a subtler effect.
COMPRESS 30–60%, DRY/WET 70–100%. On electric bass: SC HPF at 40–80 Hz to stabilize the detector against the sub. The compressor follows pick attacks on acoustic guitar particularly well.
Air-G Velvet was developed by Air-G Audio as part of a line of plugins based on the work of Airwindows.
The ButterComp2 algorithm is the work of Chris Johnson (Airwindows), whose original source code is published under an open license at github.com/airwindows. Air-G Velvet is a reimplementation of the original algorithm with its own additions: MID/SIDE architecture, sidechain filter, and user interface.
To Chris Johnson for his tireless work on Airwindows and for sharing decades of audio processing research completely openly.
Air-G Velvet v1.0 — Air-G Audio. The ButterComp2 algorithm is the work of Chris Johnson / Airwindows and is used under the terms of his open license.