When I first started recording and mixing music in my home studio, compression was something I knew absolutely nothing about. Naïvely I thought that audio compression in mixing was the same as the compression technology for the bit-rate reduction in audio and video - I even bought a book about it to learn more!
Whenever live instruments are recorded, microphones pick up the quietest and loudest parts of a performance. The difference between the softest and loudest parts are known as the dynamic range, and a compressor is a device capable of “squashing” that range. Compression raises the quietest parts and attenuates the louder parts of the audio into a preset range.
With digital technology, it is possible to create a perfectly transparent compressor and provide it as a plug-in. However, since the early 1950s external analogue equipment, including compressors, have provided many music mixes with added punch, color, and warmth.
Therefore digital plug-in manufacturers, such as Waves or Universal Audio, offer modern-day users digitized versions of the analog classics. In most cases, the digital version accurately replicates the analog version, and therefore, the analog controls and circuitry remain vital features to understand in order to get the most from them and apply them correctly to your own mixes.
Due to the nature of dynamic range reduction, a compressor is not a tool that can be set once and forgotten about. Each control available can have a dramatic effect on the output audio signal, and therefore needs to be fully understood to optimize processing results. In the first section of this article, we will look in detail at each control.
While reading through this material it may be a good idea to open your DAW’s stock compressor and play with each control to get a feel for what it does. Most compressor plug-ins provide a graphical display of each function, along with an audio trace to help visualize the processing.
The threshold control on a compressor simply sets the level at which processing is carried out on the input signal. For example, if the threshold limit is set to -15 dB any signal below -15 dB is untouched, and any signal above -15 dB is subject to processing. On some compressors threshold is adjusted by using the input gain. Setting the threshold very low will compress most of the audio signal, whereas setting threshold very high will compress a small amount of audio signal, or none at all.
Signals that exceed the threshold are subject to gain reduction according to the compressor’s ratio setting. To hear how the threshold control affects your audio signal turn it to the maximum and reduce it slowly so that you can hear its effect.
The above explanation about the threshold describes a system with a hard limit threshold and therefore describes compression with a hard-knee. A hard-knee threshold forces the compressor to immediately clamp down on signals above the threshold limit.
On the other hand, a soft-knee threshold defines two limits of threshold, between which two limits compression is applied gradually, and above which is applied consistently. On many compressors the soft-knee radius can be tightened and relaxed. The compression ratio changes and is proportional to the curvature of the knee radius.
Generally a soft-knee performs gentler compression compared to a hard-knee, and can be useful, on the mix bus for example, for gluing tracks together. Hard-knee compression is generally more aggressive than soft-knee compression.
As a signal rises above the threshold level it is detected by the compressor. The attack control determines how quickly the signal will be compressed, and ultimately the amount of overall compression. A fast attack will immediately clamp down on any signals above the threshold and reduce the gain. In contrast, a slow attack allows the initial signal to be output unprocessed (i.e. with no gain reduction), and the remainder of the signal to be processed with gain reduction.
A fast attack results in the audio signal becoming more sustained. One of the best examples to explain this is a snare drum hit. A fast attack setting on a snare track will normalise the overall volume so that the initial transient of the snare hit and the resonating sound of the drum and snare wires are brought closer together. In turn this sustains the audio signal and makes the snare drum sound “fatter”.
Attack times are expressed in milliseconds (ms). Fast attack times are generally less than 25 ms, medium attack times are around 50 ms, and slow attack times are above 100 ms.
Without the release control, signals processed by the compressor after they rise above the threshold level would continue to be processed until the level drops below the threshold. In many situations, this type of processing will cause an effect known as “pumping” and is not always desirable. The compressor is therefore fitted with a release control, which provides an additional length of time the processing will occur even after the signal drops below the threshold level.
A fast release time causes, as we mentioned above, the audio signal to “pump.” In comparison, a slow release time tends to produce audio that has smoother dynamics, and is useful for levelling audio, rather than only reducing transience and peaks. Just be careful not to cut too much of an instrument’s dynamics.
As with attack times, release times are also expressed in milliseconds (ms) but are also expressed in seconds. Fast release times are generally in the millisecond's range and slow release times in the second's range.
We already understand that the threshold level determines when a signal should be processed. A compressor processes the audio by reducing the gain. The compression ratio determines the amount that a signal is reduced while being process. A ratio of 2 to 1 halves the audio signal above the threshold. A ratio of 4 to 1 a quarter, and so on. For example, with the threshold set at -18 dB, and a ratio of 2 to 1, a signal of -10 dB will be reduced to -14 dB.
Q: With a threshold set to -20 dB, and a ratio of 4 to 1, how much would the signal of -10 dB be reduced to?
A: -17.5 db.
Low compression ratios are generally around 2 to 1. Medium compression ratios are generally around 5 to 1. High compression ratios are usually considered to be around 10 to 1. And generally, anything above a ratio of 10 to 1 would be regarded as a limiter.
The overall effect on the audio signal from compression reduces the volume level of tracks. For example, if you compress the signal in the case above (-10 dB to -14 dB) the overall volume level will go down by 4 dB. If you had initially balanced the track with the other instruments in the mix the compressed signal now needs to be 4 dB louder. You could do this by pushing the fader up 4 dB. However, the compressor offers the ability to increase the level with the built-in make-up gain control.
In turn, this increases the volume level of the quieter parts in the track too (in the above example the softer parts will also be 4dB louder). Listen out for increased noise, hand and finger movements on an acoustic guitar neck, bleed from a close snare drum microphone, or any other quieter background audio.
We have seen the standard controls of an audio compressor above, so now let’s take a little look at some more advanced features. If you are a novice compressor user, focus on the primary controls first before trying out the advanced features.
When a compressor processes audio, its detector circuits analyze the input signal and decide how much gain reduction to apply. During detection, the input signal is analyzed across the full frequency spectrum of the audio, where an average level of the full frequency is calculated.
In turn, this average level is compared to the threshold level, and the compressor applies gain reduction to the signal. As most recorded audio tracks do not have equal loudness across the frequency spectrum, the compression amount is skewed by the louder frequencies. For example, a loud kick drum will trigger compression on a mix bus, so that gain reduction is applied to all the other instruments too, therefore skewing the compression towards the kick drum hits.
To avoid skewing the compression this way, it is possible to redirect the detection input signal from another source. This is known as the side-chain. The side-chain makes it possible to:
In the kick drum example above, it is therefore possible to apply a high-pass filter to the input of the mix bus compressor, so that the low frequencies produced by the kick drum are not included in the detection circuit. Some compressors with a side-chain function also provide side-chain filtering, but if they do not the filter can easily be set up with a basic EQ and an auxiliary track.
Following the same logic, it is also possible to use a compressor as a “Ducker.” Ducking is literally the technique employed when an audio signal is attenuated in response to a second audio source.
This may best be described with an example. In the “How to Get a Killer Kick Sound” article (read here), we talk about the interaction and masking problems between the kick drum and bass guitar. If for example, we would like to attenuate the bass guitar when the kick drum is hit, we can simply use a Ducker (a compressor on the bass guitar track with a side-chain controlled by the kick drum hits) to reduce the bass guitar volume.
Some compressors provide a lookahead function. Some engineers like to think of the lookahead function as a negative attack time. Lookahead is typically done 1 ms, 5 ms, or 10 ms ahead.
In the digital world, the lookahead function is possible because the audio plug-in can “see” the audio ahead by reading the digital file. Lookahead is also possible in the analog world if an additional tape head is physically placed a few centimeters forward of the tape reading head.
The lookahead function on a compressor is particularly useful when brickwall limiting in mastering applications. The lookahead provides the compressor with advanced information about peaks and therefore can clamp down on proactively, rather than reactively.
Now that we understand the function of the input signal of the compression detector circuit, it’s time to think about how a compressor processes stereo audio.
A stereo linked compressor combines the right and left audio information and uses that to calculate the average level. The controls on the compressor affect the right and left audio information equally. The gain reduction on both sides is the same. A stereo linked compressor, when applied in moderation, tends to maintain stereo width and field.
A dual-mono compressor splits the stereo into two channels, and each channel is processed separately. A dual-mono compressor has two sets of controls, one for each channel. The gain reduction on the left-hand side will differ from the gain reduction on the right-hand side. A dual-mono compressor, even when applied in moderation, tends to pull the stereo image off-center.
On some compressors, there are auto settings, and you’re probably
wondering whether you should use these are not.
The auto release control on a compressor can combine slow and fast release times with the same setting, and is therefore helpful when dealing with loud and quiet parts of the audio. The auto release is often useful for transparent dynamic range reduction on vocal tracks, but probably not suitable for instruments like drums, bass, and guitar.
The auto make-up gain on a compressor is a useful feature as it provides a more articulate equal volume A/B comparison between the original and compressed track.
Using a compressor on a track will ultimately produce side effects (even if it’s just in the form of volume leveling, which is the function of a compressor and the reason it is being used in the first place). Sometimes the side effects add to the musicality, and sometimes not. Typically over-compressed tracks produce artifacts such as “pumping” and “breathing.” Different compressor models, through their analog modeling, can vary the tone and color of a track, again adding something that may be wanted or not. Many new digital compressors offer a very transparent compression with no side effects whatsoever.
Parallel compression, also known as New York compression, is a technique used to beef up the quieter parts of the track without losing the sense of full dynamic range. A duplicate track is made, healthily or heavily compressed, and blended back with the original signal. Typical compression ratios range from 4 to 1 to 10 to 1, with low threshold, fast attack, and slow-release settings. Gain reduction of 20 dB Is not unusual when using the parallel compression technique.
The dbx 160 and 1176 are both great compressors for parallel compression.
The parallel compression technique is often used on snare drum, kick drum, the whole drum kit, vocals, bass guitar, and even on a sub-mix bus of a group of instruments. Parallel compression is a powerful technique that can add excitement and power when applied correctly.
A compressor which is set with a very high compression ratio is known as a brickwall limiter. The purpose of the brickwall limiter is to prevent any overshoot of signal above the threshold limit. The preference for this type of limiting is to process the audio without any artifacts, such as distortion or unnatural saturation.
Brickwall limiters are typically used in one of the last stages of the mastering chain, usually before dithering. The limiter prevents the mastered audio from digital clipping and allows the overall perceived volume of the track to be maximized. However, this type of limiting is sometimes useful in other mix situations.
Multiband compression is a technique that divides the audio signal into multiple frequency bands, for example, the low, low mid, mid, and high-frequency ranges. The advantage of multiband compression is that each frequency band can be processed separately. For example, the compression ratio applied to the low frequencies could be set to 5 to 1, while the ratio applied to the high frequencies could be set to 1.5 to 1. Using this technique high volume low-frequency instruments such as the kick drum and bass guitar do not affect things like vocals, acoustic guitars, or cymbals.
After the audio signal is split and processed in each band, the signal is put back together and output as a single signal. Multiband compression is often used in the mastering chain, and on group tracks such as drums to only effect cymbal volume when the cymbals are detected in the high-frequency band. There is usually not much use for a multiband compressor on a single instrument track unless the original recording is problematic - multiband compressors can be fantastic problem solvers!
As a general rule multiband compression should be applied subtly for best results. Be extra careful matching the make-up gain in each frequency band.
There is no major difference between a compressor and limiter except for the compression ratio. Typical compression ratios on compressors are 1.5 to 1, 2 to 1, 3 to 1, 5 to 1. Whereas the compression ratio for a limiter is typically 10 to 1 or above. Having such a high compression ratio on a limiter means that the audio signal above the threshold level is dramatically reduced. For example, if a signal rises 15 dB above the threshold level, the 15 dB (with a ratio of 10 to 1) is reduced to 1.5 dB (10% of its original amount).
Limiters also tend to have very fast attack and release times so that they can clampdown peaks and transients in the audio.
The main compressor types can be categorized as follows:
Use an optical compressor when you want to even out a long stretch of audio.
Use an FET compressor on snare drums, the drum group, vocals or any tracks you want to make more dynamic and punchy.
Use a VCA Compressor on the Mix Bus or in the Mastering Chain. The VCA is also good to use when the recorded audio is very “peaky”.
The Vari-Mu compressor, due to the adaptive nature of the tube circuits, is perfect for tightening and smoothing instruments like the acoustic guitar. Just remember that it’s not good for adding aggression or punch.
The fundamental difference between these two compressor types is where the compressor detector circuit takes its input from. A feedback compressor takes its input from the compressor circuit’s output, whereas a feedback compressor takes its input from the compressor circuit’s input.
A feedforward system uses open-loop control and therefore does not react to be altered signal. On the other hand, a feedback system uses closed-loop control and continually reads the output and adjusts the control according.
Famously, the 1176 and NEVE 2254 are both feedback compressors, while the SSL G-Series Stereo Bus Compressor is feedforward. Some commercial compressors provide the option to switch between the two types, or are sometimes hybrids of both.
In the early days of audio, analog compressors were designed and used to allow radio broadcasts to limit the dynamic range of live sound. As time went on recording and mixing engineers found a use for these devices and modified the designs specifically for music applications.
This leveling amplifier (LA) was created in the early 60s by inventor, James F. Lawrence. The LA-2A, which uses a single light source and a photoresistor feedback circuit to control gain reduction, is known for its smooth sonic character and ease-of-use. The hardware and software units are controlled by one gain knob, and one peak reduction knob. This compressor is a favorite with many mixing engineers and often used on vocals, bass guitar, and the kick drum.
This classic compressor was created in the late 60s at Chicago-based company Universal Audio. The 1176 was one of the first compressors to use the newly available Field Effect Transistor (FET) technology - a technology which replaced vacuum tubes with solid-state semiconductors. This radical new technology allowed the compressor to react very quickly to audio and offered users very high attack times. Engineers love this compressor for the vitality that it imparts on tracks. This classic compressor is versatile in the studio and can be used on any tracks that need color or energy added.
Invented by David Blackmer and introduced in the mid-70s this legendary voltage controlled amplifier (VCA) compressor continues to be a staple in the audio processing world. The dbx 160 is capable of carrying out subtle compression and brick-wall limiting. The dbx 160 uses true RMS average peak levels to control its gain reduction. One benefit of RMS is that it is closer to the way that we humans hear and therefore tends to normalize the output. Amongst other things, the dbx 160 sounds fantastic on drums.
This classic compressor was created in the early 1950s by Estonian born, Soviet refugee, inventor Rein Narma. The Fairchild compressor consists of two independent limiting units and was designed with a total of 20 vacuum tubes and no less than 11 transformers. It is undoubtedly these analog components that provide the user with a highly colorful musical sound. Many professional mix engineers use the Fairchild 670, without employing any active gain reduction, to add analogue harmonics to their tracks. The Fairchild sounds great on just about everything, but especially the mix bus, vocals, and drums.
As a novice home studio owner, it is often difficult to know where to start when compressing an instrument or track. When I began mixing myself I frequently asked “the community” what were proper settings for vocals or drums. Annoyingly, the answer was “just use your ears man! Every track is different” which was never helpful as a beginner.
What I needed were hard numbers to set my compressor to, so I knew I was in the right ballpark. Hopefully, the table below will provide you with useful information to use on your own tracks. The chances that any of the settings will be perfect for your track as slim to none, however we’ve all got to start somewhere.
|Instrument||Ratio||Attack||Release||Threshold||Make Up Gain|
|Vocals I||1.5 to 1||20ms||40ms||Set to reduce gain by 2 to 3db||2dB|
|Vocals II||2 to 1||5ms||20ms||Set to reduce gain by 2 to 3db||2dB|
|Vocals III||5 to 1||1ms||40ms||Set to reduce gain by 2 to 3db||2dB|
|Bass I||3 to 1||2ms||100ms||Set to reduce gain by 3 to 4db||3dB|
|Bass II||3 to 1||10ms||40ms||Set to reduce gain by 4 to 5db||4dB|
|Bass III||4 to 1||100ms||25ms||Set to reduce gain by 5 to 6db||5dB|
|Guitar I||2 to 1||25ms||200ms||Set to reduce gain by 1db||1db|
|Guitar II||3 to 1||30ms||200ms||Set to reduce gain by 1 to 2db||2dB|
|Kick Drum I||3 to 1||4ms||200ms||Set to reduce gain by 3 to 5db||3db|
|Kick Drum II||4 to 1||4ms||200ms||Set to reduce gain by 4 to 6db||4dB|
|Kick Drum II||4 to 1||50ms||200ms||Set to reduce gain by 5 to 7db||5dB|
|Snare Drum I||4 to 1||3ms||50ms||Set to reduce gain by 4 to 6db||4dB|
|Snare Drum II||6 to 1||10ms||100ms||Set to reduce gain by 6 to 8db||6dB|
|Snare Drum III||8 to 1||15ms||250ms||Set to reduce gain by 8 to 10db||8dB|
|Overheads I||2 to 1||1ms||250ms||Set to reduce gain by 2 to 5db||2dB|
|Overheads II||3 to 1||10ms||500ms||Set to reduce gain by 4 to 7db||4dB|
|Overheads III||4 to 1||30ms||1s||Set to reduce gain by 6 to 9db||6dB|
Table 1: Instrument Compressor Setting Starting Points