Compression Explained
COMPRESSION
There is only one studio effect that is more useful than the compressor, so useful that it is nearly always incorporated into the mixing console itself - EQ. Equalization is nearly always necessary, but you can learn to apply EQ by instinct. Dynamic control is harder to get the hang of and use successfully, so while you are instinctively EQing, you can be applying some of the dynamics techniques outlined here. Before looking at what a compressor does, as well as the noise gate, let's first of all consider why dynamic control is necessary in recording. After all, if we don't need compressors strapped to the sides of our heads, why do we need them when we record sound?
But wait... In fact, the human ear DOES have its own built in compression system. It's a bit more compact than the units we normally mount in our nineteen inch racks, but it is nonetheless very effective. The ear basically consists of three sections which are called the outer ear, middle ear and inner ear. The outer ear is basically that flap you have on the side of your head and the connecting passage down to the ear drum which picks up the sound. The inner ear is the transducer of the system converting vibrations, which are now travelling through fluid, into nerve impulses. There is a problem getting the sound vibrations from the eardrum into the fluid in the cochlea - the organ of the inner ear - because sound doesn't very much like to change from travelling in one medium to another. If there was no middle ear to make a correction, as much as 30 decibels of level would be wasted (this is apparently OK for some animals, but not for humans). The middle ear therefore has a mechanical amplifier consisting of three bones which act as levers, making up the loss. This is where the ear's compressor has its effect. When the incoming sound gets too loud, as it might if you go to a concert by the local jack hammer quartet, a muscle around the three bones tightens and skews the levers reducing their efficiency, and therefore the level entering the inner ear. It's very clever and gives good protection against continuous loud sounds.
With the aid of its internal compressor, the ear can manage a pretty wide range of sounds. In quiet enough surroundings you can indeed hear a proverbial pin drop, and probably sounds even lower in level than that. At the other extreme, you probably wouldn't like to get too close to the take off runway at LAX, but if you approached it gradually you would find that the noise was at first loud, then uncomfortably loud, and then painfully loud. Between the extremes of level from the quietest sound the ear can hear to the loudest the ear can handle there is a range of approximately 120dB. It may come as a surprise in this age of technological marvels that few types of sound equipment can come even close to this. An analog tape recorder can manage, on a good day, a dynamic range of about 65dB between its noise floor and the highest level it can record with acceptably low distortion. 16 bit digital equipment can in theory have a signal to noise ratio of 96dB. Even this is 24dB less than the ear's range. '24dB less' means that if you matched the volume so that the quietest sound your digital system could produce was the same as the quietest sound the ear could hear, then the loudest sound available from the digital system would be at a sound pressure level only one sixteenth of the ear's maximum level. The upshot of all this is that any sound we want to record has to be squeezed through a system which has a much narrower dynamic range than the ear. The quietest sounds will need to be boosted if they are not going to be lost among low level noise, and the loudest sounds will need to be kept under control. This means that we need compression.
There is another aspect to dynamic range, apart from the fact that even modern equipment is not as 'hi-fi' as the ear. Do we actually need such a wide dynamic range when we play back a recording at home? If the loud parts of the music are too loud, then might not the neighbours become just a bit upset? And if the quiet parts are too quiet, will you still be able to hear them above the background noise inevitable in almost any domestic setting? The only answer is to control the dynamic range to make sure that it isn't too much, either for the equipment to handle or for the listening environment. This doesn't always mean using a compressor; sensible mixing and use of the faders is one of the best ways of making sure that the dynamic range of a recording is as you want it.
If one function of the compressor is to control dynamic range, then a completely separate function is its use as an effect. Take 'effect' to mean changing a sound from its natural state into something intentionally different and artificial, either to improve it or simply make it more interesting. As will be seen later, the noise gate has 'corrective' and 'creative' uses too. And when we put the two together then we will really find out the power of dynamic control.
COMPRESSING VOCALS
One of the main uses of compression as an effect is recording vocals. Trained singers can sing at a consistent level and keep the difference between loud and soft passages within acceptable limits. Untrained singers usually don't have the same degree of breath control. Indeed, singing according to the rule book involves breathing from the diaphragm rather than the chest, and this does not form part of the vocal style of many of the most popular vocalists, so we need to compress. Compression is used to bring down the highest peaks, above the threshold level, leaving the lower levels just as they were. After that the level is restored so that the peaks are the same level as they were to start with, but the overall dynamic range is reduced. The result is a much more controlled sound.
COMPRESSOR CONTROLS
Most compressors have pretty much the same controls:
1. Threshold sets the level where compression starts to take effect. Sounds below the threshold pass through unaltered and only sounds above the threshold are compressed.
2. Ratio sets the degree of compression above the threshold level. A ratio of 2:1 represents mild compression and means that when the incoming level (that is, the level above the threshold) rises by 10dB, the outgoing level will only rise by 5dB. Ratios of up to 5:1 are regularly used for vocals and other instruments, and can pass by unnoticed by the listener if the other controls are set properly. Higher ratios are used for more serious limiting, where the level needs more severe control. Ratios of 10:1 and higher are nearly always noticeable to the listener.
3. Attack is measured in milliseconds and determines the time taken for the compressor to start working once the signal has passed the threshold level. Why have Attack? Because sometimes if you leave just a hint of attack (say, for instance, a snare drum hit, or a powerful downstroke on an acoustic guitar) then it sounds a bit more realistic. Removing too much attack can sound a touch fake.
4. Release sets the length of time it takes for the compressor to return to its normal state once the signal has gone back below the threshold.
5. Gain is provided because compression always reduces the peak level: the more compression, the lower the level of the outgoing signal. This control is sometimes referred to as 'make-up' gain because it makes up the level that is inevitably lost during the compression process.
Of these five controls (there may be more on a more sophisticated compressor), Threshold, Ratio, Attack and Release are the most important. With these you can either improve a sound beyond all recognition or you can ruin it beyond salvation. Thinking about vocals and individual instruments for the moment, it's useful to decide what ratio to use just by listening, then start on the other controls. It's not the kind of procedure where you can set one control, fine tune it and then go on to the next. All the controls of the compressor seem to interact with each other and need careful balancing. Having fairly arbitrarily set the ratio, next up is the threshold level. This is where the important gain reduction meter on the compressor comes into play. The gain reduction meter tells you what the compressor is doing to the signal from moment to moment and gives you a visual check on what you are hearing. A gain reduction of from 6 to 12dB is usually OK. Above this much compression you might hear that the sound is being squeezed too much (although it can be useful for an effect, as opposed to simply controlling the sound). Anything much below 6dB compression and you'll have some difficulty hearing the difference. An important point which often isn't mentioned is that at some occasions during the track, there should be no gain reduction (as indicated by the meter) at all. If the only time the meter reads zero dB gain reduction is when the instrument isn't playing, then when it does start you'll get a 'crack' as the compressor takes effect. Setting the threshold to a lower value normally gives more compression, but setting it to a lower value than necessary just gives you the undesirable side effects.
There is a balancing act between the Ratio and Threshold controls to give the right amount of compression, according to what your ears perceive, and there is also a balancing act between the Attack and Release controls and the signal, to reduce the potential obtrusiveness of the compressor. The principal reason why compressors have Attack and Release controls is so that you can make the compression come and go unnoticed. The settings have to be judged against the attack and decay envelope of the signal. Drums for example naturally have a fast attack and decay, so if the attack and release settings are too slow, the compression will not start quickly enough and the initial transient will get through uncompressed, and after the drum sound has gone you will hear the compressor gradually reducing the amount of gain reduction as it returns to normal. This last effect is known as "breathing" or "pumping", and occurs when you can hear background noise changing in level unmasked by the signal. If the Attack and Release controls are set properly, the signal should hide any level changes in the background. The best way to set attack and release is to match the attack and decay characteristics of the signal itself. That way you won't get more pumping than necessary.
Having said how to set the attack and release times correctly, let's explore how they can be set incorrectly, but with a useful effect. One use of the compressor is to make drum sounds more punchy. This is done, just as mentioned above, by setting a slow attack. The initial transient of the drum then gets through before the compressor clamps down on the remainder of the waveform. This is a very useful technique for guitars as well. Curiously enough, the compressor is actually working to expand the dynamic range around the time of the initial transient. What a crazy world! The Release control, as well as having an effect on how much pumping you will hear, can also change the amount of compression going on. In fact the Release control has as much effect on that as the Ratio and Threshold controls does.
Here's an example: Years ago, at an audio exhibition, I heard about a salesman that was demonstrating a high-end compressor. "That's 30dB compression right there", he was saying, which is the same as saying 30dB gain reduction, which is a lot, "Can you hear it working?". Listeners couldn't hear any pumping at all, nor could they hear the characteristic compressor sound which you would certainly expect to hear from a unit working that hard. Looking a little closer at the gain reduction meter, it was right at the end of the scale as expected, but more importantly it wasn't moving. This is a crucial point about compression. Unless the gain reduction meter is moving then no compression is taking place. The demonstrator had craftily set a very long release time and virtually all that the compressor was doing was lowering the gain on a steady basis. Big deal... To get any use out of the compressor, not only do you have to see the LEDs of the gain reduction meter light up, you also have to see them move. The faster they move up and down, the more effective compression you are getting. (And the more you are likely to hear pumping). Needless to say, it's unrealistic to expect everyone who needs a compressor to purchase a $5000 unit, so one that we recommend is the FMR Audio RNC Compressor. At $175, it's truly a remarkable unit. It's know for being very transparent and doing it's job with no audible artifacts. Don't just take out word for it. Do the research. You'll see they're quite impressive and loved by many!
The other important control you will always find on a compressor is the Stereo Link switch. This is quickly explained: If you are compressing a stereo signal, the levels in the two channels will be unequal and will change from moment to moment. Unless something is done the amount of compression will be different in the two channels, making the stereo image shift from one speaker to the other according to which channel is being compressed the most. To avoid this, the Stereo Link switch mixes the control signals for the gain elements together so that each channel is compressed to the same extent.
There is only one studio effect that is more useful than the compressor, so useful that it is nearly always incorporated into the mixing console itself - EQ. Equalization is nearly always necessary, but you can learn to apply EQ by instinct. Dynamic control is harder to get the hang of and use successfully, so while you are instinctively EQing, you can be applying some of the dynamics techniques outlined here. Before looking at what a compressor does, as well as the noise gate, let's first of all consider why dynamic control is necessary in recording. After all, if we don't need compressors strapped to the sides of our heads, why do we need them when we record sound?
But wait... In fact, the human ear DOES have its own built in compression system. It's a bit more compact than the units we normally mount in our nineteen inch racks, but it is nonetheless very effective. The ear basically consists of three sections which are called the outer ear, middle ear and inner ear. The outer ear is basically that flap you have on the side of your head and the connecting passage down to the ear drum which picks up the sound. The inner ear is the transducer of the system converting vibrations, which are now travelling through fluid, into nerve impulses. There is a problem getting the sound vibrations from the eardrum into the fluid in the cochlea - the organ of the inner ear - because sound doesn't very much like to change from travelling in one medium to another. If there was no middle ear to make a correction, as much as 30 decibels of level would be wasted (this is apparently OK for some animals, but not for humans). The middle ear therefore has a mechanical amplifier consisting of three bones which act as levers, making up the loss. This is where the ear's compressor has its effect. When the incoming sound gets too loud, as it might if you go to a concert by the local jack hammer quartet, a muscle around the three bones tightens and skews the levers reducing their efficiency, and therefore the level entering the inner ear. It's very clever and gives good protection against continuous loud sounds.
With the aid of its internal compressor, the ear can manage a pretty wide range of sounds. In quiet enough surroundings you can indeed hear a proverbial pin drop, and probably sounds even lower in level than that. At the other extreme, you probably wouldn't like to get too close to the take off runway at LAX, but if you approached it gradually you would find that the noise was at first loud, then uncomfortably loud, and then painfully loud. Between the extremes of level from the quietest sound the ear can hear to the loudest the ear can handle there is a range of approximately 120dB. It may come as a surprise in this age of technological marvels that few types of sound equipment can come even close to this. An analog tape recorder can manage, on a good day, a dynamic range of about 65dB between its noise floor and the highest level it can record with acceptably low distortion. 16 bit digital equipment can in theory have a signal to noise ratio of 96dB. Even this is 24dB less than the ear's range. '24dB less' means that if you matched the volume so that the quietest sound your digital system could produce was the same as the quietest sound the ear could hear, then the loudest sound available from the digital system would be at a sound pressure level only one sixteenth of the ear's maximum level. The upshot of all this is that any sound we want to record has to be squeezed through a system which has a much narrower dynamic range than the ear. The quietest sounds will need to be boosted if they are not going to be lost among low level noise, and the loudest sounds will need to be kept under control. This means that we need compression.
There is another aspect to dynamic range, apart from the fact that even modern equipment is not as 'hi-fi' as the ear. Do we actually need such a wide dynamic range when we play back a recording at home? If the loud parts of the music are too loud, then might not the neighbours become just a bit upset? And if the quiet parts are too quiet, will you still be able to hear them above the background noise inevitable in almost any domestic setting? The only answer is to control the dynamic range to make sure that it isn't too much, either for the equipment to handle or for the listening environment. This doesn't always mean using a compressor; sensible mixing and use of the faders is one of the best ways of making sure that the dynamic range of a recording is as you want it.
If one function of the compressor is to control dynamic range, then a completely separate function is its use as an effect. Take 'effect' to mean changing a sound from its natural state into something intentionally different and artificial, either to improve it or simply make it more interesting. As will be seen later, the noise gate has 'corrective' and 'creative' uses too. And when we put the two together then we will really find out the power of dynamic control.
COMPRESSING VOCALS
One of the main uses of compression as an effect is recording vocals. Trained singers can sing at a consistent level and keep the difference between loud and soft passages within acceptable limits. Untrained singers usually don't have the same degree of breath control. Indeed, singing according to the rule book involves breathing from the diaphragm rather than the chest, and this does not form part of the vocal style of many of the most popular vocalists, so we need to compress. Compression is used to bring down the highest peaks, above the threshold level, leaving the lower levels just as they were. After that the level is restored so that the peaks are the same level as they were to start with, but the overall dynamic range is reduced. The result is a much more controlled sound.
COMPRESSOR CONTROLS
Most compressors have pretty much the same controls:
1. Threshold sets the level where compression starts to take effect. Sounds below the threshold pass through unaltered and only sounds above the threshold are compressed.
2. Ratio sets the degree of compression above the threshold level. A ratio of 2:1 represents mild compression and means that when the incoming level (that is, the level above the threshold) rises by 10dB, the outgoing level will only rise by 5dB. Ratios of up to 5:1 are regularly used for vocals and other instruments, and can pass by unnoticed by the listener if the other controls are set properly. Higher ratios are used for more serious limiting, where the level needs more severe control. Ratios of 10:1 and higher are nearly always noticeable to the listener.
3. Attack is measured in milliseconds and determines the time taken for the compressor to start working once the signal has passed the threshold level. Why have Attack? Because sometimes if you leave just a hint of attack (say, for instance, a snare drum hit, or a powerful downstroke on an acoustic guitar) then it sounds a bit more realistic. Removing too much attack can sound a touch fake.
4. Release sets the length of time it takes for the compressor to return to its normal state once the signal has gone back below the threshold.
5. Gain is provided because compression always reduces the peak level: the more compression, the lower the level of the outgoing signal. This control is sometimes referred to as 'make-up' gain because it makes up the level that is inevitably lost during the compression process.
Of these five controls (there may be more on a more sophisticated compressor), Threshold, Ratio, Attack and Release are the most important. With these you can either improve a sound beyond all recognition or you can ruin it beyond salvation. Thinking about vocals and individual instruments for the moment, it's useful to decide what ratio to use just by listening, then start on the other controls. It's not the kind of procedure where you can set one control, fine tune it and then go on to the next. All the controls of the compressor seem to interact with each other and need careful balancing. Having fairly arbitrarily set the ratio, next up is the threshold level. This is where the important gain reduction meter on the compressor comes into play. The gain reduction meter tells you what the compressor is doing to the signal from moment to moment and gives you a visual check on what you are hearing. A gain reduction of from 6 to 12dB is usually OK. Above this much compression you might hear that the sound is being squeezed too much (although it can be useful for an effect, as opposed to simply controlling the sound). Anything much below 6dB compression and you'll have some difficulty hearing the difference. An important point which often isn't mentioned is that at some occasions during the track, there should be no gain reduction (as indicated by the meter) at all. If the only time the meter reads zero dB gain reduction is when the instrument isn't playing, then when it does start you'll get a 'crack' as the compressor takes effect. Setting the threshold to a lower value normally gives more compression, but setting it to a lower value than necessary just gives you the undesirable side effects.
There is a balancing act between the Ratio and Threshold controls to give the right amount of compression, according to what your ears perceive, and there is also a balancing act between the Attack and Release controls and the signal, to reduce the potential obtrusiveness of the compressor. The principal reason why compressors have Attack and Release controls is so that you can make the compression come and go unnoticed. The settings have to be judged against the attack and decay envelope of the signal. Drums for example naturally have a fast attack and decay, so if the attack and release settings are too slow, the compression will not start quickly enough and the initial transient will get through uncompressed, and after the drum sound has gone you will hear the compressor gradually reducing the amount of gain reduction as it returns to normal. This last effect is known as "breathing" or "pumping", and occurs when you can hear background noise changing in level unmasked by the signal. If the Attack and Release controls are set properly, the signal should hide any level changes in the background. The best way to set attack and release is to match the attack and decay characteristics of the signal itself. That way you won't get more pumping than necessary.
Having said how to set the attack and release times correctly, let's explore how they can be set incorrectly, but with a useful effect. One use of the compressor is to make drum sounds more punchy. This is done, just as mentioned above, by setting a slow attack. The initial transient of the drum then gets through before the compressor clamps down on the remainder of the waveform. This is a very useful technique for guitars as well. Curiously enough, the compressor is actually working to expand the dynamic range around the time of the initial transient. What a crazy world! The Release control, as well as having an effect on how much pumping you will hear, can also change the amount of compression going on. In fact the Release control has as much effect on that as the Ratio and Threshold controls does.
Here's an example: Years ago, at an audio exhibition, I heard about a salesman that was demonstrating a high-end compressor. "That's 30dB compression right there", he was saying, which is the same as saying 30dB gain reduction, which is a lot, "Can you hear it working?". Listeners couldn't hear any pumping at all, nor could they hear the characteristic compressor sound which you would certainly expect to hear from a unit working that hard. Looking a little closer at the gain reduction meter, it was right at the end of the scale as expected, but more importantly it wasn't moving. This is a crucial point about compression. Unless the gain reduction meter is moving then no compression is taking place. The demonstrator had craftily set a very long release time and virtually all that the compressor was doing was lowering the gain on a steady basis. Big deal... To get any use out of the compressor, not only do you have to see the LEDs of the gain reduction meter light up, you also have to see them move. The faster they move up and down, the more effective compression you are getting. (And the more you are likely to hear pumping). Needless to say, it's unrealistic to expect everyone who needs a compressor to purchase a $5000 unit, so one that we recommend is the FMR Audio RNC Compressor. At $175, it's truly a remarkable unit. It's know for being very transparent and doing it's job with no audible artifacts. Don't just take out word for it. Do the research. You'll see they're quite impressive and loved by many!
The other important control you will always find on a compressor is the Stereo Link switch. This is quickly explained: If you are compressing a stereo signal, the levels in the two channels will be unequal and will change from moment to moment. Unless something is done the amount of compression will be different in the two channels, making the stereo image shift from one speaker to the other according to which channel is being compressed the most. To avoid this, the Stereo Link switch mixes the control signals for the gain elements together so that each channel is compressed to the same extent.
