Dynamic Range Compression (DRC), also referred to simply as “compression,” is a procedure in which an audio signal’s dynamic range is lessened. Compression can be used in many different situations, including broadcasting, live sound reinforcement, and sound recording to helps control audio levels. The type of device used to control these levels is referred to as an audio compressor.
It is easiest to think of audio compression as automated volume control. There are two different forms of compression: downward compression and upward compression. In downward compression, any part of the signal that goes over a volume threshold will be pulled down, while sounds that are quieter than the threshold remain untouched. In upward compression, quieter sounds are pushed up, while louder sounds remain untouched.
With compression, we are able to lower the audio signal’s dynamic range. There are many reasons for compressing a signal, for example, to improve a signal’s audibility or change a signal’s aesthetics.
When an environment is very noisy, that noise in the background can overwhelm sounds that are quieter. For example, if you were listening to your stereo in the car without compression, the sound of the road would likely overpower your quieter music. This is especially true if you are trying to listen at a comfortable level. To help give you clarity at a tolerable listening level with both quiet and loud signals, stereo manufacturers use compression.
Compression helps to pull down the loud sounds while keeping the quieter sounds constant. In doing so, it allows you to raise your volume without the loud sounds being to loud. On the opposite end of the spectrum, many people use what are called audio expanders, which increase an audio signal’s dynamic range.
The gain or level of a signal is reduced by an audio compressor when the amplitude of that signal goes above a given threshold. The amount of gain that is reduced is determined by the ratio. When you use a ratio of 2:1—for example, whenever your input signal gets 2 dB over the given threshold—the signal that comes out will only be 1dB. Essentially, you have reduced your overall signal level by 2 dB in this case. Another example would be if your signal went over a threshold by 4 dB with an 8:1 ratio, there would be a level reduction of 4 dB.
Check out these settings below to get a better grasp on this:
- Ratio is 3:1
- Threshold = -5 dB
- Input = -2 dB (3 dB above the threshold)
- Output = -4 dB (1 dB above the threshold)
When a signal enters an audio compressor, that signal is split into two halves. One copy is sent to a side-chain path and one side is sent to a variable gain amplifier. The side-chain path signal uses a control circuit to calculate the amount of gain reduction that is required. Whatever the gain-reduction amount that the amplifier requests, the control circuit will output. This type of design, which is used on the majority of modern compressors, is known as a feed-forward design.
Back in the day when early compressors hit the market, they made use of feedback layouts. The control circuit took the signal it was fed after it went through the amplifier.
To reduce the gain of a signal, a compressor uses a variable gain amplifier. There are a variety of different methods and tools that can do this, which each have their own pros and cons. The ‘variable-µ’ configuration uses vacuum tubes where the gain is altered by grid-to-cathode voltage. Another technique is known as a voltage-controlled amplifier, which bases its gain reduction on the input signal.
Audio Compressors that use light-sensitive resistors (LDRs) and small lamps (Electroluminescent or LED panels) are known as optical compressors. Many believe that optical compressors are the smoothest around. This is because the attack of the incoming signal is softened by the resistors. Lastly, there are other technologies that make use of a diode bridge and field-effect transistors.
Implementing compression in digital audio is a bit different. To implement compression via dedicated workstations or digital audio editors, we have to use digital signal processing techniques. There are many digital compressor manufacturers that try to emulate some of the analog methods listed above.
Features of an Audio Compressor
The signal is reduced when it goes above a specified level, known as the threshold. The threshold typically uses dBs. If you were to set your threshold very low (somewhere around -50dB for example), a bigger portion of your signal would be compressed or treated relative to a higher threshold setting (around -5dB, for example).
The input/output ratio for signals that exceed the threshold is determined by something known as the ratio. If you have a 4:1 ratio, for example, any part of the signal that exceeds the threshold by 4db will actually output the compressor with only 1dB. Note that 60:1, which is more commonly referred to as infinity:1, is one of the highest compression settings. Any signal that goes above the threshold will be brought down. We call this brick wall limiting.
Attack and Release
You can choose how quickly your compressor reacts by using the attack control. The period in which the compressor increases gain reduction to get to the level that was determined using the ratio is known as the ‘attack phase’. The period where the compressor decreases gain reduction to get to the level that was determined using the ratio is known as the ‘release phase’. Essentially, the compressor will let go of the signal once it has fallen below the threshold.
We can determine the length of these periods by the required change gain reduction and the rate of change. Audio Compressors typically have controls that are labeled ‘attack’ and ‘release’ (usually in milliseconds), which help make the overall operation far more intuitive. We can think of this as the time that it will take for a compressor to change the amount gain, measured in decibels. While this gain change is typically decided by the manufacturer of the compressor, it is usually 10dB.
If a compressor references 10dB for its time constants and you set the attack to 1ms, this means that the time it takes for the gain reduction to move from 0dB to 10dB is 1 ms. It will take 2 ms for the gain reduction to move from 0dB to 20dB, and so on and so forth.
While many compressors allow users to adjust the attack and release times, there are many compressors that have set attack and release times. These attack and release times are pre-determined by the circuit design, meaning a user will not have the ability to adjust them. Attack and release times may also be program dependent or automatic. Consequently, the attack and release times will vary depending on the level of the input signal. This may actually alter the overall sound or character of your signal. Some compressors add characteristics that are quite obvious depending on the loudness of the incoming signal.
Soft and Hard Knees
The hard knee and soft knee are two choices that some compressors give us. The response curve on a compressor has a bend to it, and by choosing hard or soft knee, you can either make your edge rounded or sharp. Soft knees act a lot slower in character by increasing the compression ratio gradually as the level increases or moves towards the user’s compression ratio. The audible change that comes from compression isn’t as noticeable when you use a soft knee. This is especially true if you are using really high ratios to compress.
Peak vs. RMS
Compressors that use peak sensing will respond to the incoming signal more instantaneously. Peak sensing reduces the gain very quickly, resulting in much tighter control and compression that is far more evident. Sometimes this can actually cause distortion. This is also the reason why many compressors include RMS, which is an averaging function that hits the signal just before it goes over the given threshold. RMS basically helps to keep audio compression sounding much more musical. It is similar to how we perceive loudness.
To apply the same amount of gain reduction to both the left and right channels, a compressor must be in stereo linking mode. Stereo linking helps stop a stereo image from shifting in the field. Shifting can occur if both the left and right channels are compressed independently but the audio content on one side is louder than the other. For example, you might have a louder signal on the right side of a drum set because the floor tom is over there.
To achieve stereo linking, we can use one of two methods. The compressor can sum the left and right channels to a mono signal when it hits the compressor’s input. This means that only the left channel of the compressor will be functional. On the other hand, the compressor could reduce both the left and right channels independently of each other while applying the greatest amount of gain reduction to both of those channels. In this case, you could use different settings on each of your channels if you don’t want the same amount of compression on one side vs. the next.
Some compressors have a parameter called make-up gain, which allows you to add a bit of level or gain to your output signal so that it comes out of the compressor at a level that is most optimal.
Using a Compressor
There are many different ways an engineer might choose to use a compressor. For starters, an engineer might use a compressor in a subtle fashion to help reduce the overall dynamic range of a specific signal during the recording phase. In this way, it will come into the recording device with an optimal dynamic range. On the other hand, an engineer could also crank a compressor so that the settings are more intense, effectively changing the character of the incoming signal or instrument.
If an engineer was looking to reduce the dynamic range of a signal without changing the overall character of the signal, they could use a very low compression ratio and a relatively high threshold. This would help to compress the signal most of the time, though in a very slight manner.
On the other hand, let’s say an engineer wanted to soften the overall attack of a snare drum. They would use a higher threshold with a fast attack and a fast release. If that engineer wanted to accentuate the attack or transient of the snare drum, they would use a slower attack. In doing so, the compressor wouldn’t squash down the initial transient.
To successfully apply attack and release times, it is best if an engineer has knowledge of different instruments and their characteristics. Of course, compression can also be used to pull a sound towards a compressed “middle” when softer passages or sections come around. Because loud passages are pulled down, softer passages are inevitably boosted or pulled up.
Limiting is a bit like compressing. In fact, the process of limiting is the same as compression; they are just perceived a bit differently. Limiters only have higher ratios than compressors, and they generally use attack times that are relatively fast. Ask most engineers and they will tell you that a ratio of 10:1 or above is considered limiting. As we know, though, there are no rules in mixing. Soft and hard limiting exist too, and they are categorized by degree differences. A limiter that uses a higher ratio and faster attack and release times would be referred to as a hard limiter.
We call high ratio limiters with fast attack times brick wall limiters. One would use a brick wall limiter if they never want the level of a signal to get above the threshold in amplitude. Brick wall limiters use ratios of up to infinity:1 and down to 20:1. Brick wall limiters are typically characterized by their harsh and un-musical sounds thanks to their momentary methods. This is why most people will only use them as precautionary devices. People in the broadcasting world use them to sculpt the sound.
Limiters can be found in a wide variety of consumer electronics. For example, the Automatic Volume Limiter System (AVLS) is used by Sony in a range of different products, including the PlayStation.
When you compare limiting and clipping, you can see that clipping introduces a pretty large amount of distortion. Limiting, on the other hand, introduces a very small amount of distortion, though it still helps to keep the signal from going above the threshold.
Side-chaining is the process of using a signal level of another signal or input, or a version of the original input that has been equalized, to control the original signal’s compression. Some side-chains use keys for external inputs. When the external input becomes stronger, the amount of compression intensifies. One example of side-chaining is when disc jockeys speak into the microphone and the volume of their music lowers.
The microphone level of the DJ converts to a line level before it is routed to the sidechain input on the stereo compressor. The level of the music, which is also routed into the stereo compressor, now ducks whenever the DJ talks into the microphone, effectively reducing the music’s volume. This process is known as ducking.
Some side-chain compressors come with equalizer controls. These controls can be used to reduce specific chosen frequencies in the range. One big example of this is a de-esser, which typically acts on the higher end above 6kHz to help get rid of unwanted sibilance. You can create a compressor that is frequency dependent if you have a regular compressor. Simply take a signal that is boosted around 6-7kHz and feed it into the side-chain’s input.
These high frequencies tend to be boosted in pre-emphasized media, such as FM radio and phonographs, which can cause problems and overdrive the overall signal. This is why we choose to de-ess them.
One other popular use of side-chaining can be found on the production side. This is when producers or engineers want to keep the integrity and loudness of a bass while getting it out of the way of a kick drum.
If a stereo compressor doesn’t have a side-chain, it can still be used as a mono compressor side-chain. The sidechain signal or key can be routed to the main or first stereo compressor input, and that compressor’s output can then be routed back into the other compressor channel.
Another wonderful audio compression technique used by many engineers is performed with a compressor in the parallel path of a given signal. Parallel compression, otherwise known as New York Compression, can provide users with all of the benefits and advantages of compression without the nasty artifacts. If the ratio on a compressor is low in relativity, the sound of the compressor will be quite natural. However, if a compressor is using a compression ratio that is higher in relativity, this can result in audible artifacts.
These artifacts can be utilized in the parallel realm. Tons of mixers use this type of compression as an effect to enhance their music. When you combine a compressor signal with a linear signal and cut back on the output gain coming from the compressed signal, you can create enhanced, low-level details while retaining your transient content. This is because the compressor will only add to the overall gain of the signal at very low levels.
Parallel compression is most beneficial when highly transient content is compressed, as you can still maintain the dynamic nature of the signal even with all of the gain reduction that is occurring on the compression chain.
Multiband compression is quite unique since it has the ability to act on frequency bands in an independent fashion. One of the biggest advantages of multiband compression when you compare it to regular full-band compression is that level changes coming from a specific frequency band, such as the heavy, low-end of a bass or kick drum, won’t cause any pumping in the compressor or unwanted gain changes that are audible.
Multiband compressors use a specific number of crossover filters or bandpass filters and split the incoming signal through them. Depending on the multiband compressor that you choose to use, you may be able to edit or adjust the crossover frequencies. You can then independently adjust the attack, release, ratio, and threshold of each independent band, essentially acting as multiple compressors. On the output side, the frequency bands are then combined. Typically, multiband compressors use added limiters at the end of their circuits to make sure that the combined signal doesn’t have any unwarranted peaks.
Multiband compressor DSP emulations or software plugins can be fairly complicated in nature. Many of these also require a good amount of computer power thanks to their many bands.
While multiband compressors are very popular among mastering engineers, as they are wonderful mastering tools, more and more mix engineers are beginning to include them in their plug-in setups. Both AM and FM radio stations use hardware-style multiband compressors for the on-air radio station signal. This helps to increase the overall perceived loudness of the radio station without over-modulating the signal.
When it comes to commercial competition, having the loudest sound is typically seen as beneficial. Of course, when adjusting the parameters of a multiband compressor for a radio station, you do need a good pair of ears and some artistic or stylistic sense. The reason is that a multiband compressor can have an EQ effect on a signal as the spectral balance between the bands is constantly changing. This can dynamically modify the frequency response on-air.
Programmable radio output processing is a further development with this approach. Here, the multiband compressor parameters will change automatically depending on the time of the day, program block, or sound coming through.
One audio compression technique that is used in the mixing and recording world is serial compression. By using two different compressors in the same input chain, you can achieve serial compression. The first compressor is used to control the overall dynamic range of the incoming signal, while the next compressor will clamp down on the louder peaks in a more aggressive manner.
When you look at combination devices, such as compressor-limiters, you’ll typically find that the standard internal signal routing works in this way. For example, a combo device might use an RMS compressor that controls the overall gain in a general sense, and that RMS compressor will be followed by a peak-sensing limiter that provides overload protection with fast-acting parameters.
If this technique is used properly, it is possible to use very heavy compression while still retaining a fairly natural sound overall. If you’re working with an extremely dynamic vocal or acoustic guitar, for example, serial compression is a wonderful choice.
Common Uses for Compression
Compression is often used to create perceived loudness without increasing the peak amplitude of the signal. It becomes possible to increase the overall level of a signal without going over the dynamic limits of the reproduction device or signal, as compressors will compress the loudest peak signal. When compression is applied alongside a boost in gain, you get a net effect of louder sounds remaining the same and quiet sounds being boosted.
This is why compression is typically used in professional audio systems that are found in retail stores, restaurants, and other public spaces. It allows the volume to stay fairly consistent while music is played, even if that music is at a relatively low volume. Essentially, it helps the music to stay louder than ambient or background noise.
One can use compression to increase the power amplifier’s average output gain to 100% from 50% while reducing the dynamic range. This helps to cut down on the number of amplifiers that are required in paging and evacuation systems while retaining clarity—perfect for circumstances that are noisy.
Music producers often use compression to help make certain vocal or instrumental compressors more consistent. By doing so, they can help a sound sit nicely within the mix of other instruments or vocals while also giving it a constant, center-of-attention quality so that the listener can enjoy every nuance of it. When you listen to a pop or rock song, you are usually listening to compressed vocals. This helps create clarity so that they stand out when they are surrounded by a slew of other instruments.
Music producers also use compression to create effects on certain instruments instead of boosting the overall loudness of the instrument. A compressor, for example, could extend the tail on the end of a drum or cymbal track. You may want this instead of the quick decay that comes naturally with drums. We often compress guitar sounds as well as it helps to create a sound that is sustained and full.
The majority of compression devices that are capable of performing the above tasks also have the ability to reduce the level of a chosen audio source as a separate audio source hits a certain amplitude.
By feeding a compressor with an equalizer that is set up with specific frequencies, you can reduce sibilant sounds. Only specific frequencies will act on the compressor. Sibilance, even if the sound levels are pretty low, can actually cause distortion. We use this process, known as de-essing, to combat that distortion.
Amateur radio stations that make use of SSB modulation use compression in their voice communications. The signal of that particular station will then be easier to read for another station that is far away. This can help that radio station’s signal stand out in front of other stations. If a radio station is in competition to talk to a DX station, amateur stations will often use compression. It is perfect for amateur station pileups.
The amplitude of an SSB signal is dependent on the modulation level. Because of this, the average amplitude of a signal and the average of the transmitted power would be much stronger if compression was not utilized. Speech compressors are typically built into SSB transceivers in most modern amateur radio stations.
Land mobile radio also uses compression. This can be used in remote dispatch consoles to help transmit walkie-talkie audio.
Broadcasting is another very common place where compression is used. It helps to boost the perceived volume of an outputted signal while reducing the overall dynamic range. This reduced dynamic range is necessary as narrow range broadcast signals have the ability to accommodate it. There are legal limits for broadcasters in most countries when it comes to the instantaneous peak volumes that they are allowed to broadcast. Typically, these broadcasters will use on-air hardware in their chain that is permanently inserted. This hardware is usually multiband compression.
Compressors are used to boost the perceived volume in broadcasting so that one station might seem louder than the next. This can be very beneficial if a station seems louder than another when someone flips from one dial to the next. The station that is louder and more compressed might impress the listener more than the other, especially at a given volume. When AM rock stations became more competitive in the 1960s, many of them began using this technique.
Because AM radio had a very poor dynamic range anyway, AM broadcasters didn’t have any qualms about this heavy compression technique. To boost the highs and lows and achieve a sound that was tight and punchy, radio stations would use the Gates Sta-level technique. This was paired with increased overall volume.
DJs from the 60s who would talk and shout over their music also used this type of heavy compression. Essentially, a DJ could mix their voice into their music instead of it feeling like they were shouting over top of it. Of course, DJs had to use a loud voice when they were delivering their patter, as it was the only way that the compressor would act. This idea of over-emphasizing words is now often parodied. These over-emphasizing DJs, referred to as “pukers” in the business, were the foundation for the ability to talk within the music rather than on top of it.
When the mid 60s rolled around and rock music became far more prevalent on the FM sphere, many radio stations began using the legendary CBS Volumanx/Audimax compression rig. The main reason was that it was an expander, working to lift any soft volume.This meant that it wouldn’t expand a signal that was unmodulated or boost the hissing noise floor, which many original units did. Unfortunately, if the compression was driven too aggressively, it would create a nasty pumping or sucking effect.
Compression works against FM’s wider dynamic range to help deliver modulation or a volume level that is more constant. This was a strong point for FM in the comparison between it and AM. This was one reason why classical and easy-listening stations of the 70s, otherwise known as “album rock stations,” strayed away from using heavy compression. In fact, classical stations barely used any type of compression. This is why many classical listeners, especially when in the car, would have to turn their volume up and down to combat the ambient noise that came from being on the road.
You can listen to the same recording on different mediums, from AM to FM to CD, and you’ll get a very different dynamic experience with each.
Mixing engineers, mastering engineers, and record companies began increasing the overall volume of commercial albums when CDs came about. At the beginning, engineers would only increase the volume until the loudest point of the record reached the limit. Then when they began using heavier compression and limiting on their masters, they had the ability to maximize the level of audio in the digital world. In fact, compression algorithms have been specifically engineered to do this.
One critic described the idea of hard clipping or hard limiting a signal as “dogshit” because it affects the tone and timbre of music in a negative way. Many engineers use the term “loudness wars” to describe the effort to increase loudness.
The majority of television commercials are heavily compressed to a dynamic range that typically does not exceed 3dB. This helps commercials stay within permissible limits while attaining perceived loudness at the near-maximum. Television viewers and listeners have been a part of this chronic problem for years now. Television commercials utilize the same concept as radio broadcasters as it helps them to create a perceived loudness in relativity. The problem is that television programs, especially older movies or television shows that have really soft dialogue, are not very compressed like television commercials.
When a commercial pops on, it could potentially blow the viewer out of his or her seat with its incredible loudness. Because older movies and television shows have very low amounts of audio energy, solving this problem is fairly difficult. Unfortunately, you can’t do much with an audio expander either, though some try to use them to even out the levels. This is why there is a disparity when it comes to volume levels, even when you flip across the dial from TV station to TV station.
Compressors can also be used with a signal transmission to help reduce the overall dynamic range before expanding it later down the line. This can help reduce unwanted effects for channels that don’t have a wide dynamic range.
When it comes to music production, engineers generally try to avoid regular amplitude peaks causing gain reduction for the rest of the mix. This is known as gain pumping. Of course, if you listen to modern hip-hop and dance music, this is actually a phenomenon that is used purposefully. In utilizing this method, one can create a rhythmic pulse within the song.
Hearing aids also use compressors to help bring down the level of audio that is received by the person wearing them. Binaural compression is used to help the patient hear from the specific direction that the sound is coming from.
Software Audio Player Compressors
Compression is often implemented by support plugins for software audio players to increase the overall volume of audio tracks. Considering the fact that music can be highly variable from classic to EDM, for example, this can help to increase the volume that we perceive as we move from song to song. This compression technique can be used to improve the listenability of audio when it is being played in environments with tons of ambient noise or on speakers that are poor quality.
This software is also used in home-based audio mastering and micro-broadcasting. To obtain playback that is volume-compressed when played on devices like computer-based audio players, these files might need to be outputted as mp3s, wavs, or other types of audio formats.
- Understanding compressors and compression
- Example and explanation of over compression
- Article on Optical Compressors
- Gain Control Devices, Side Chains, Audio Amplifiers
- Universal Audio
- Sound On Sound, December 2000. Paul White. Advanced Compression Techniques
NOTE: Incase the download above is not working, the Archive.org version is available here