Sidechaining and keying
We often cover the basics of dynamics processing in these pages, focusing mainly on simple applications of compression, limiting, gating, and expansion, but advanced processing functions can be achieved using dynamics processors, through the use of techniques known as sidechaining and keying. This month we’ll explain the concepts behind key inputs and sidechains, and explore applications for their uses. Most of these ideas apply to hardware and software processing, though the execution varies somewhat. While you’re reading through this, please remember that the key input or sidechain to a dynamics processor is separate from the audio path.
Example of automated key operation, using MOTU Digital Performer. Watching the Detectors All dynamics processors contain a control circuit, often referred to as a detector. This is the hardware circuit (or software emulation thereof ) that does the thinking for the device. The detector analyzes the audio signal and—based upon the way you configure controls—applies the requisite gain control. [Editor’s note: the sound of a compressor is all about the type of detector circuitry the designer has employed. Optical, Vari-Mu, and VCA compressors possess very different gain control characteristics, resulting in distinct sonic characters—all of which are eminently useful. That’s a discussion for another time.]
Under most circumstances, the detector is listening to the same audio signal that is being processed. This sounds deceptively simple: After all, why wouldn’t the detector be controlled by the same audio that it’s processing? The answer is, a lot of advanced dynamic effects can be created when the detector reacts to a signal other than the audio signal it is processing. To accomplish this, a hardware compressor must have a separate key input or a sidechain path consisting of a dedicated input and output. You can route anything you want to this input and use it to take control over the detection process. For example, you can route a kick drum to the detector and let the kick drum govern gain reduction on a vocal; see the block diagrams shown in Figures 1A and 1B. Figure 1A (right) shows audio in the compressor being split and routed to the audio path and the detector. Figure 1B shows a vocal in the audio path, but a kick drum is routed to the detector’s key input. Every time the kick drum is hit, the compressor acts on the vocal.
Fig. 1A. Audio sent through the compressor is split and routed to the audio path and the detector. The beauty of using plug-ins is that many of them feature a key input, even if that plug-in is emulating a hardware device that did not have a key input. For example, the original Universal Audio 1176 limiter did not have a key input, but the Bomb Factory BF76 1176 emulation does, letting you accomplish processing that was impossible with the original hardware unit.
Many plug-ins feature a key input, even if that plug-in is emulating a hardware device that did not have a key input. For example, the original Universal Audio 1176 limiter did not have a key input, but the Bomb Factory BF76 1176 emulation does (see upper left corner), letting you accomplish processing that was impossible with the original hardware unit. Key Control Enabling key operation simply requires feeding any signal you’d like into the key input. Let’s take as an example the commercial “donut”: Radio and TV commercials typically start with a few seconds of music, and then a voiceover comes in to deliver the sales pitch. In the olden days, an engineer would manually lower the music when the voiceover started so that the music would not compete with the message. When the voiceover ended, the music would briefly return to the original level before fading out.
This process can be automated by using the voiceover to control the gain of a compressor inserted on the music track. The routing for this process is demonstrated in the screen shot of Digital Performer’s mixer window. In this window, the track on the left (highlighted red) is the voiceover. The track on the right (highlighted blue) is the music.
MOTU’s dynamics processor is inserted on the music track and is set to Compressor. The drop-down menu for the Control Signal is open. You’ll see Input at the top of the menu and a list of buses underneath. When the Control Signal is set to Input, the dynamics processor behaves as you’d expect: It acts based upon what the music is doing—i.e. when the music track gets louder, compression increases. However we have set the Control Signal to Bus 1.
The Voiceover track has a send on Bus 1; the send knob is turned up and the send is set Prefader so that the key signal is independent of the vocal fader. This routing enables the compression on the music track to be “triggered” or “keyed” from the voice track. When the voice starts, it feeds audio to the Control input of the dynamics plug-in, causing the music track to be compressed.
When the voiceover ends, the signal at the Control input ceases and the music track returns to normal volume. This process is known as a ducker. Note that once the Control Signal is set to Bus 1, the compressor will not function unless there is a signal present on Bus 1. A similar technique can be used in a karaoke situation or for a restaurant paging systems in which the announcer’s voice ducks the background music.
Fig. 1B. In this diagram, the kick drum is the control signal, keying compression on vocals. There are other uses for a ducker, one of which is a compression trick from back in the metal days. If you have a song with distorted rhythm guitars and you want to keep them “in yer face,” use the lead vocal to key compression on the guitar tracks. Every time the vocal enters, the guitars are reduced in level. You’ll need to set the parameters of the compressor so that the effect is subtle: Try medium to fast attack and quick release times so that as soon as the voice starts, the guitars duck and as soon as the voice ends the guitars come back to normal. Start with a ratio of 2:1 or 3:1; you’ll need to play around with the threshold to achieve just a few dB of compression so that the effect is not obvious. The nice thing about doing this in a DAW (as opposed to the old analog days) is that you can easily route the vocal to key compressors on multiple guitar tracks without the hassle of physically splitting the signal and connecting a lot of patch cables etc. Simply set the key inputs of the compressors to the same bus as the vocal send.
I use the same technique all the time with delay effects on lead vocal, live and in the studio. Ducking the delay while the vocalist is singing keeps the vocal in front and maintains clarity. When the vocal ends, the delay comes up, making it more audible.
Fig. 2. Here, the vocal is fed into the compressor, and an EQ is patched into the sidechain. Filter Your Mouth Sidechaining is a similar concept, but involves sending the detector signal out to an external processor, altering it, and then returning it back to the detector input. Patching an EQ into the sidechain is very common, and can be used to create a de-esser. A de-esser is actually a compressor that has been made sensitive or “tuned” to sibilant frequencies. This process is achieved by applying an EQ to the signal before it returns back to the detector. In the block diagram shown in Figure 2, the vocal is fed into the compressor, and an EQ is patched into the sidechain. (Note that the compressor’s sidechain switch must be engaged or the EQ is not applied.) I usually dump out all of the frequencies below around 3 kHz and apply a severe boost in the upper mids (anywhere from 4 to 8 kHz, depending upon the singer. You’ll have to experiment), making the compressor very sensitive to sibilance. If you set the threshold, attack and ratio controls carefully, the compressor will act on “s” sounds but other sounds will not trigger compression.
Sidechain filtering can also be used to avoid excessive compression due to a high content of low frequencies in a mix. Low frequency sounds carry a lot of energy in a mix and they can trigger compression that causes audible side effects. For example, you might notice that every time the kick drum hits, the lead vocal gets sucked down. The cure is to remove some of the low frequencies from the compressor’s sidechain (say, everything below 200 Hz). Filtering the bottom end stops the compressor from kicking in every time a kick drum is hit, but remember—you have not filtered the audio path.
“Smarter” hardware gates such as the Drawmer DS404 feature front panel controls for low- and high-frequency filters. Using both filters simultaneously allows you to build a bandpass filter that removes some of the sounds that you do not want opening the gate. Keys to the Kingdom Similar concepts can be applied to expander/gates. As with their compression counterparts, hardware gates will feature a key or trigger input jack, usually with an associated switch on the front panel. Plug-ins will offer a key on/off button along with a drop-down menu that allows you to choose a key input. This could be a physical input on your audio interface but more likely will be a bus that will receive a signal from elsewhere in the session.
Think of a gate as a door that opens or closes based on the strength of the signal at the doorway. If the signal is strong enough the door is pushed open, but the key or trigger input of a gate is like an electric door latch. That latch opens or closes based on a remote signal—regardless of the strength of the audio that is attempting to pass through the doorway. A gate that provides a key input allows you to use a secondary sound to open the “doorway.” Let’s say you insert a gate on a synth bass track but route the kick drum to the gate’s key input. The synth bass itself will not open and close the gate—the kick drum will control the gate, allowing the synth to be heard or not. This can be used for some interesting effects (usually in the studio) where the synth is musically tightened up to the kick drum hits. In fact such a technique could be used to change the rhythm of the synth bass so that it precisely matches that of the kick drum. Swap the synth bass for a test tone generator tuned between 50 and 80 Hz, and you’ll have a TR808 kick sound. A similar technique can be used to tighten up “gang” vocals by keying them from the lead vocal track.
Live engineers on major tours have been known to use contact pickups or triggers on each drum to key that drum’s gate instead of using the signal from the microphone to open the gate. Let’s say you have a mic on a snare drum and you are attempting to gate the mic. Depending upon placement of that microphone and the player’s touch on the kit, audio from other components of the kit such as toms, cymbals and kick drum may leak into the microphone, causing false triggers that open the gate even when the snare is not hit. If we add a contact pickup or trigger to the snare drum and route it to the gate’s key input, the gate opens and closes much more reliably because the trigger is in physical contact with the drum and is much less subject to leakage (though some contact pickups may be sensitive to vibration). This also means that softer hits on the snare can open the gate reliably so that grace note-style hits are not muted. As a bonus, we could split the signal from the trigger or contact pickup and send it to the trigger input of a drum module—allowing us to layer a sampled snare with the real snare or possibly record the performance as MIDI data.
Gates that have a sidechain (or a sidechain filter) can be especially useful because they let you filter unwanted sound from the sidechain—preventing them from opening the gate and effectively making the processor more sensitive to the sounds you do want opening the gate. A great application for a sidechain filter would be when you have significant kick drum leakage into a snare drum track, and the kick drum is causing the snare gate to open. By filtering the low frequencies from the sidechain, the gate will “hear” less of the kick drum and respond more to the snare. Remember that this filter applies to the sidechain and not to the main audio path. Most plug-in gates include a sidechain filter (Waves’ Renaissance Channel for example), so take advantage of it. “Smarter” hardware gates such as the Drawmer DS404 feature front panel controls for low- and high-frequency filters. Using both filters simultaneously allows you to build a bandpass filter that removes some of the sounds you do not want opening the gate. You’ll also typically find a sidechain “listen” switch that enables you to temporarily hear the filtered sidechain signal via the audio outputs. This is extremely useful in tuning the filters to pass only the signal you are trying to gate.
Take a Look Ahead One feature that is exclusive to software gates (and compressors) is the look-ahead function. Here’s how it works: A gate or compressor can only act upon a signal when it reaches the input of the processor. In some cases, a transient such as a snare hit might fly right past the gate before the gate can respond—so that hit will be muted. Look-ahead allows the gate’s sidechain to hear the signal before it hits the gate’s audio input (did someone say “cheating”?), allowing it to open just before the sound reaches the gate’s audio input. MOTU’s MasterWorks Gate is great for this because you can set the look-ahead in milliseconds from 0 to 20. I find that setting it to 1 or 2 milliseconds is just enough that the gate can be set tightly enough to cut leakage but not chop off the leading edge of a snare or tom hit. A similar function on a compressor plug-in can help catch transients that are faster than the comp’s attack time.
It’s worth mentioning that if you are using hardware compression or gating, be careful about processing during recording. If a gate inadvertently removes audio that you wanted (that grace note on a snare, for example), it cannot be recovered. This is much less of an issue in DAWs because plug-in effects are almost always non-destructive—in other words, they are applied in the monitor path and can be removed without harm to the original audio file.
Steve La Cerra is an independent audio engineer based in New York. In addition to being an Electronic Musician contributor, he mixes front-of-house for Blue Öyster Cult and teaches audio at Mercy College Dobbs Ferry campus.