When managing vocal extremes, one can always count on a compressor and/or limiter being an important tool of the trade. Although one of the most common audio processes, compression/limiting is often one of the most misunderstood. Many first-time users don’t fully understand how to deploy compression so that it doesn’t sound as if it’s being used. As a result, they shoot to normalize a vocal part, and end up just crushing the dynamics to dust.
To start from the very beginning, a compressor/limiter is essentially an automatic volume control that consists of four basic parameters: ratio, threshold, attack, and release. Ratio narrows the dynamic range of the performance by reducing the gain once it reaches a predetermined signal threshold. A 3:1 ratio, for example, works like this: When the input signal increases in gain by 3dB, the output from the compressor will increase by only 1dB. This reduction of the output peak volume thereby “compresses” the overall dynamic range of the performance.
Threshold determines at which dB level the compressor engages. Any basic compressor will have what are called hard knee and soft knee threshold settings. Hard-knee compression is when the amplitude of the input signal is a straight line until it reaches the dialed-in threshold setting, at which point the “angle” changes abruptly. The higher the ratio is set, the harder the angle.
Conversely, a soft-knee setting allows a wide progressive adjustment as the threshold is approached, therefore creating a curve correction rather than a hard angle. As you can imagine, the soft knee offers a smoother transition, which, depending on the situation, may be preferred. Attack determines the speed at which the compression is applied after the threshold has been reached. Release controls the time the compressor takes to return to pre-threshold settings.
Take note that a short attack, quick release, high ratio, and hard-knee threshold can create an undesirable pumping and breathing effect. Setting your release time to longer than the source signal’s natural decay will prevent pumping, and smooth out the tone, making the vocal seem more sustained. The only potential problem with this is if your vocal track has a lot of noise floor, as the noise will be sustained, as well.
There are three common places to apply compression: in the record path, as an insert on a mix channel, and on an output bus. During recording, compression applied to a voice is often positioned in the signal chain directly after the mic preamp. When the compressor is on, the output peaks will be reduced, but the output level, including noise, is increased. Using a compressor first in the chain gives other effects a more consistent signal to process, so it’s good practice to place the compressor directly after the mic preamp.
Stereo compression is often used on the output bus as a final process to increase the apparent loudness and presence of an overall mix. Multiple passes of compression in low ratios (such as at the record stage, and later on during the mix) will result in an overall smoothing quality, and will reduce the potential for undesirable byproducts (such as those pumping and breathing anomalies).
Multiband compressors act the same as non-multiband compressors, but instead adjust volumes to both ratio settings and predetermined frequency bands. For this reason, multiband compressors are particularly useful when processing the human voice, as it’s rare to encounter a human voice that results in a perfectly flat frequency curve. While multiband compressors have historically been regarded as primarily broadcast and mastering tools, their inclusion in many DSP packages have led to frequent use at the mixing stage. A de-esser—which can reduce undesirable sibilance (most notably on “s,” “sh,” and “zh” sounds) in the 2kHz to 10kHz range—is a commonly recognized example of a multiband compressor. This tool can be especially helpful in keeping harsh high-end sounds at bay (even from cymbals, percussion, and fuzzed-out guitar tracks).
With an incredibly loud vocal performance, compression alone may not be sufficient, and limiting may be needed to help tame the meters. Compressors and limiters behave identically, but limiters typically handle compression ratios of 10:1 and upwards. Limiting is frequently used as a failsafe with a “do not pass” threshold level—with the processor often inserted in the bus output in environments with legal decibel limitations, or to protect live sound gear—as nothing is more damaging to a performance than a huge moment of overloading distortion.
However, the outcome of over-limiting can equally be crude and jarring, as a sudden gain reduction disrupts the listener’s experience. Extreme limiting also affects the point of perspective, because a sudden reduction results in the source being perceived as being further away spatially. One can certainly argue that the sound of limiting is better than the sound of overload distortion, but, even so, use limiters with extreme caution, and apply them moderately—unless you want to squash the life out of your track.
Noise gates and expanders conduct an inverse process of compressing and limiting—they increase dynamic range by attenuating signals that fall below a user-defined threshold. A noise gate, for example, lets through only those signals that are above the gate’s threshold. This is great for ensuring that breaths and background noise don’t compromise a vocal performance. However, if you set the threshold too high, the gate might also clamp down part of the vocal, causing chopped syllables or weird stutters. Plosives (boomy pops caused by “p” and “b” sounds) can also be battled with a gate. If you’re dealing with a plosive, set the attack of the gate moderately fast, so that it attenuates the sound, but doesn’t “click” when opening. Try 1ms as a good default setting, as it will preserve intelligibility of the words being sung without making the sound to choppy. Make sure to keep the decay time relatively long (approximately 0.5 seconds), so that you don’t snip off the end of any words, and keep the threshold set as low as you possibly can, so as to keep as much of the dynamic performance intact (although unwanted noise levels shouldn’t be able to open the gate).
While a noise gate attenuates signals swift as a guillotine blade, an expander offers a gradual attenuation. This is a good tool to use when a vocalist takes breaths, say, in the middle of a silent passage (as opposed to just before a lyric line, where using a gate would make more sense), or if noise is apparent when the singer isn’t singing (such as during a guitar solo).
Perfecting dynamic range is a matter of understanding your source signal, and working the parameters akin to a sculptor using clay. There are no hard and fast rules. When choosing your tools and planning out your mixing strategy, you can’t just look at the vocal performance itself—you must first look at the whole picture. And you can’t simply watch your dials and meters when applying any of these tools. Before touching a single button or cranking any knobs, you must ask yourself: Is the voice one part of a dense musical arrangement, or a delicate and exposed soundscape? What is the dynamic range of the vocal performance vis-à-vis the dynamic range of the surrounding instrumentation? And then listen, tweak, wash, and repeat.
STEREO MIKING FOR A DYNAMIC VOCAL PERFORMANCE
If you have found your vocal dynamic range has been extreme in the past, and that limiting, while necessary, has altered its natural sense of space in the mix, it may be wise to track your vocals with two matching mics in a stereo pattern. To do this, place both mics equidistant from the source—your mouth—and apply a high-threshold compression setting to mic one, and a noise gate to mic two. One microphone will take off where the other has left off, yet both will have manageable dynamic ranges. Make sure each signal is panned in the same pattern, and match everything in both signal paths to preserve uniformity of your sound.