An excerpt from the Mixing Audio: Concepts, Practices and Tools by Roey Izhaki, available here.
Following compressors, gates are perhaps the second most common dynamic range processors in mixing. Gates are also called noise-gates – a name that implies their traditional usage as noise eliminators. In the past, tapes and analog equipment were the main contributors to added hiss.
Nowadays, digital technology tends to allow much cleaner recordings. Still, in project studios unwanted noise on a recording might be the outcome of ground-loops, a computer fan, or airborne traffic.
Noise, however, is not the only thing gates are employed to reduce. The rumble produced by the floor tom when the rest of the drum kit is played is often gated. The hi-hats spill on a snare track and the headphone spill on a vocal track are just two more examples.
In addition to their traditional role, gates are also used for more creative tasks, like tightening drums, adding punch or applying dynamic movement.
Inside a gate. The vertical chain shows the main side-chain stages, and the controls link to each stage.
Gates affect signals below the threshold – these are either attenuated or muted. Signals above the threshold pass through unaffected, unless some attack is applied.
A gate only cares whether the signal is above or below the threshold; a gate is said to be closed when the signal is below the threshold and open when the signal is above it.
Figure 1 illustrates the function of a gate on a snare. Threshold settings on a gate might seem straightforward – we set the threshold below the wanted signals and above the unwanted signals. In practice, this affair can be tricky since not always a certain threshold meets both criteria.
Figure 1: Gate threshold function.
For example, the snare in Figure 3 lost some of its attack and much of its decay. We could retain more of both by lowering the threshold, but this would cause false triggering by the hi-hats. Solutions for this common issue are discussed later on. In the meantime we should note:
Lower threshold settings on a gate are often sought after as they help in retaining more of the natural attack and decay. The original signal before gating involves a single snare hit and a few hi-hat spikes we wish to eliminate.
The threshold is set above the hi-hat peaks, so the gate would only open once the snare hits. The gate would close as the snare hit drops below the threshold. After gating, only the loud portion of the snare remains.
To allow quick gate opening once the signal exceeds the threshold, the level detection on most gates is based on peak-sensing.
Level fluctuations are more profound with peak-sensing than with RMS. While fluctuating in level, signals may cross the threshold in both directions many times over a short period of time. This causes rapid opening and closing of the gate, which produces a type of distortion called chattering (see Figure 4a).
One way to overcome this is by having two thresholds – one above which the gate opens, another below which the gate closes. Having two threshold controls would be cumbersome since adjustments to one will call for adjustments to the other. Instead of providing two controls, gates offer a single threshold and a control called hysteresis.
Figure 2 shows hysteresis control. (a) is a gate with single threshold and no hysteresis control. Chattering is introduced due to the rapid opening and closing of the gate, which are triggered by quick level changes. (b) is gate with hysteresis control. The gate only opens as the signal overshoots the opening threshold and only closes as the signal drops below the closing threshold. Level variations between the two thresholds do not cause toggling of the gate state, thus no chattering occurs.
Figure 2: Hysteresis control on a gate.
The threshold is the opening threshold, and the hysteresis determines how many dB below it the closing threshold is set. For example, with the threshold set to −20 dB and hysteresis to 5 dB, the closing threshold would pose at −25 dB. Figure 4b illustrates this.
Many gates do not offer hysteresis as an adjustable control, but have a built-in setting fixed between 4 and 10 dB. When hysteresis control is given, these figures provide a good starting point.
Range, or depth, defines the amount of gain applied on signals below the threshold. A range of −10 dB means that signals below the threshold are attenuated by 10 dB.
Often signals below the threshold are considered muted, although in practice this perceived muting is due to heavy attenuation with the typical range of −80 dB.
Figure 3 shows the transfer function of a gate. (a) is a gate with 0 dB range: both below and above the threshold the input–output ratio is at unity gain, and the gate would have no effect on the signal. (b) is gate with −20 dB range: all signals below the threshold are simply attenuated by 20 dB. For example, an input signal at −40 dB will leave the gate at −60 dB.
Figure 3:The transfer function of a gate.
Meanwhile, (c) is a gate with −80 dB range, which effectively mutes all the signals below the threshold. (d) shows again a −80 dB range, but with an extended output scale that reveals what happens below the limits of the standard scale. We can see clearer here that an input signal at −40 dB will leave the gate at −120 dB.
Figure 4 demonstrates the effect of different range settings. When the range is set to −5 dB, signals below the threshold are attenuated, but still heard. With large range, such as −80 dB, signals below the threshold become inaudible.
Figure 4: The range function of a gate.
The range value can be expressed in either positive or negative values. Depending on the manufacturer, a gain attenuation of 40 dB might be expressed with a range of 40 or −40 dB.
This book uses the negative notation. Small range denotes little effect (e.g., −5 dB), while large range denotes more effect (e.g., −80 dB). Generally speaking, large range settings are more common in mixing. However, sometimes it is only a gentle attenuation we’re after.
One example would be gating vocals to reduce breath noises between phrases – removing breaths altogether is often perceived as artificial, so these are often only attenuated. Another example involves reducing the room reverb captured on a recording.