Dynamics Signal Processing

ADVANCED RECORDING PRIMER

CHAPTER 3 - PART 2, BY ROBERT DENNIS

DYNAMICS SIGNAL PROCESSING

History and Uses:

Dynamics Signal Processing is automatic gain changes that change the level relationship between loud and soft passages. Dynamic Processing comes under the categories of Compression, Limiting, Expansion & Gating. It is used for reducing the dynamic range to record with a system that has dynamic range restraints (like analog). It is also used for raising apparent volume, reducing noise and leakage, preventing overload, modifying attack or sustain, reducing sibilance or pops, keeping a consistent level, reducing the need for other processing.

Basic Explanations & Basic Terms:

Dynamic processing is one of the most effective tools that a recording engineer has to make sure all of those important parts that have been recorded will actually be heard in the final product. The term "Dynamic Range" refers to the level difference between the loudest musical passage and the softest musical passage. Dynamic Processing devices change this level difference.

TYPES OF PROCESSING

Compression

Compression means that any signal that tries to exceed a set "threshold" level will not be allowed to increase as much. In other words the compressor turns down the maximum passages so they won't be as loud. With the threshold control set to maximum, no passages will be turned down; as the threshold control is brought down some peaks of the music will exceed the threshold and be turned down. Compression is used to even out the volume of the music so that all passages are more even.

.Expansion

The term "expand" means that any signal the falls bellow the threshold level will be made to go lower in level than it would without expansion. In other words expansion is just the opposite of compression. One use of expansion is to "undo" the over-use of compression. Another use would be to make low level sounds (like mechanical sounds of instruments) less noticeable.

Figure 16a - Effect Of Compression & Expansion

Limiting

The term '"1imit" means that the signal will not be allowed to exceed the threshold level at all. Limiters are used to prevent distortion by setting the threshold control just below the distortion level. The limiter then prevents any peak from getting to the level where it would distort.

Figure 16b - Limiting Compared To Compression

Gate

The term "gate" refers to turning off the signal when it falls below the threshold level. A range control is included so that signal can just be turned down rather than completely off. Gates are used to get rid of noise and leakage when the instrument is not playing. This is done by setting the threshold level lower than the level of the signal but higher than the level of the noise or leakage.

Figure 16c - Gating Compared to Expansion

DYNAMICS PROCESSING CONTROLS

Threshold

The threshold control adjusts the level where the dynamic processor will start to work. In a compressor or limiter when the loud passages at the input exceed the threshold level set, the unit will turn down these loud passages. In an expander or gate, the unit will turn down any incoming signals that are below the threshold level. The threshold is usually adjustable by a control marked "threshold." Reducing the threshold level means that more peaks of the signal will trigger the compressor to turn down the gain; increasing the threshold level means that an expander or gate will turn down more low-level signals.

Ratio

The ratio control determines how much the signals that are being compressed or expanded will be turned down.

If a compressor has a 2:1 ratio, the compressor will turn down the gain so that if the input signal is 2dB above the threshold level, the output increases only one dB. If the input signal is 4 dB above the threshold, the gain will be turned down so that the output only rises 2 dB above the threshold (a 2 to 1 ratio). At a 4:1 ratio the input signal has to be 4 dB above the threshold for the output to increase 1 dB. When the ratio control is set to 10:1 or more, the compressor is called a limiter because the unit is effectively preventing the peak levels from increasing any significant degree above the threshold level.

Expanders also have ratios. An expansion ratio of 1:2 means that if the input signal drops 1 dB below the threshold level and the gain will be reduced so that the output is 2 dB below the threshold level. If an expander has a very high expansion ratio, it becomes a gate.

Gain Reduction Meters

You can tell "how much" compression or expansion you are using by reading the gain reduction meter. The meter will read how many dB that the signal is being turned down by the compression or expansion. With the gain reduction meter reading "-6" the output is being turned down dB or to half the level on that particular passage.

Figure 18a - Gain Reduction During Compression

Figure 18b - Gain Reduction During Expansion

Attack & Release Times - The attack time is how fast the dynamics processor will react to a signal crossing the threshold level, going up. In a compressor it is the time it takes the compressor to reduce gain on a high-level passage. On an expander, it is the time that the expander takes to restore full gain after the audio level comes up after a low level passage.

The release time is how fast the dynamics processor will react to a signal crossing the threshold level, going down. In a compressor, it is the time it takes the unit to restore gain after the high-level passage is over with. In an expander, it is the time the expander takes to turn down a low level passage (below the threshold level).

Very generally speaking, fast attack times are good. Release times should be adjusted for the frequency and how percussive the signal is. Compressors for bass signals must not be set to a very short release time or the gain will be changed within one cycle, causing distortion. Generally speaking, the fastest release time that sounds natural is the best for both compression and expansion.

Makeup Gain (Output Level) - The makeup gain is a level control at the output of the compressor. The compressor reduces signal strength of the louder passages. After this compression, the peak level of the signal is less. This makeup gain control allows the compressed signal to be turned up so that its peaks are as high as they were before compression. This action of "making up" gain increases the lower level passages of the signal.

Knee - How abruptly or gradually the compression ratio is applied to the signal when the signal is near the set threshold level. A Hard Knee means that the full compression ratio is applied to any signal above the threshold level. A Soft Knee means that a mild compression ratio is applied to signals approaching the threshold level and that the ratio is increased as the signal rises above the threshold level; the full compression ratio is applied to signals well above the threshold level. The soft knee makes the gain reduction less obvious - sort of like turning down the volume of a stereo slowly rather than abruptly. Soft knee is used in applications where you are using a compressor to even out volume changes in an instrument. DBX, a well-known brand of compressors, trademarked their soft knee function with the name Overeasy.



The graphs to the right are the traditional way of showing the input/output characteristics of hard knee and soft knee compressors, and the graphs resemble "knees" (hence the name). The above graph shows how the output signal would be different in these two types of compressors.

Sidechain – A sidechain is the control circuit of a compressor. Usually the compressor's sidechain simply receives the input signal into the compressor [and most explanations of how a compressor works assume that the input signal also goes to the sidechain].

On many compressors you can insert a different signal into the sidechain by plugging into the "sidechain" or "key" input. (on some compressors you also have to push a button labeled "sidechain" or "key"). When a different signal is put into the sidechain, the compressor operates according to the signal fed to the sidechain input rather than to the main input. In other words the main signal (going into the regular input and output) is changed by the level variations of the signal plugged into the sidechain input.

RMS (Average) Detecting - There are two ways you could define the "level" of a signal. One way would be to use the absolute peak level of the waveform; the second way would be to use the "average" level of the waveform. In percussive instruments, such as drums, there is a huge difference between the peak level and the average level. In other instruments, such as a flute or violin, the two levels are much closer. Electronic circuits overload according to the "peak" level but your ear responds to the "average" level. If you are using a compressor to help prevent overload (say of a digital recording) you would want the compressor to respond to the peak level. If you are using a compressor to make an instrument more even in volume, you would want the compressor to respond to the "average" level. the term RMS is an acronym for the mathematical process of determining a signal's effective average level [the "M" stands for "mean" - another word for average]. Many compressors allow the user to chose between average and peak detecting.

Auto Release - Auto Release is variable release time feature in compressors and helps make the signal's volume even to the ear. With auto release activated, the compressor's release time is longer for peaks going well above the threshold and shorter for peaks going slightly above the threshold.

Range - The Range Control is a expander or gate control. It sets the maximum amount of dB that the expander or gate will reduce gain on lower level passages.

Breathing - Breathing is hearing the gain restore after a compressor works, with you hearing the noise or leakage increase between notes. Generally breathing comes from the release time being too long when hefty amounts of gain reduction are being used.

Uses:

MAKING UP FOR DYNAMIC RANGE LIMITATIONS

A mild amount of compression while recording can keep the lower-level passages above the noise of analog recording. This would be under the category of general compression. General compression also serves to make a higher average or "apparent level." The settings for this general compression are as follows.

General Compression Settings
Threshold	About 6 db. below higher peaks
Ratio	2:1 to 4:1
Attack	Medium (40-60 ms.)
Release	Auto or Medium (100 - 150 ms)
Knee	soft
Note: Attack time parameter may be reduced (all the way to 0) if percussive quality in enhanced too much but this is more effected by the knee parameter (which may have several soft knee settings). Release time should be reduced if breathing is present.

During mixdown in analog recording the noise of the medium may be reduced with a gate or expander.

ADJUSTING THE AMOUNT OF ATTACK AND RELEASE

The transient attack of percussive instruments (like a sharp snare drum) occurs in the first 3-4 milliseconds of the sound. You can prove this to yourself by mixing a sharp snare sound with a delay of the sound (say with delay of 10 milliseconds). If you did this you would hear a "double-hit" on the snare. This fact helps you set the attack and release parameters of compressors to change the attack of such an instrument.

Dulling Out Attack - Sometimes sounds are "too percussive" to our ears. To dull out the attack you would want to use a very fast attack time (0 ms) a low ratio (like 3/1) and a soft knee to affect the slope of the transient sound. It is often helpful to increase the release time to between 10 and 25 ms, which makes the "processing" you are doing with the compressor less obvious. Try a slightly higher amount of gain reduction, say 10 dB. Increasing the ratio will give more dulling but will make the instrument sound unnaturally "clamped" or "restricted."

Increasing the attack - Sometimes we would like more attack than is present in the instrument or the mix. To increase the attack, use an attack time of about 5 to 10 ms. The idea is that you want the body of the sound brought down by the compressor but not the transient. My experience is that a medium ratio (6/1) and a hard knee works best in these circumstances. The attack time can be increased to taste and the best attack time varies with the instrument.

GETTING MORE APPARENT VOLUME

Compression - The "General Compression" settings described under "Making Up For DP Limitations" above will gently give you more apparent level without the processing becoming obvious. Many "mastering" settings have these parameters with the ratio on the low side of the range given (2:1).

Limiting - In today's world of digital media (like the compact disk) the recorded level is determined by the peak level of the signal. Our ears, however, perceive loudness by the average level of the signal. If we could cleanly sheer off the last 6 dB of the percussive attacks, we could raise the level of the signal up 6 dB to get more average level (and loudness) but still retain much of the percussive quality of the sound. This could be a technique used in mixing percussive instruments or even in the mastering of the overall mix. The result we would want is shown in the diagram.

The parameter settings to achieve this include a very high ratio (say 20/1 or more), hard knee, very fast attack (0 ms if possible) and very fast release (5 ms). You adjust the threshold so that you get 6 dB or so gain reduction on the peaks.

REDUCING THE NEED FOR OTHER PROCESSING

The "General" Compression settings tend to reduce differences in how loud different "bands" of frequencies are, reducing the need for equalization. General compression also reduces the level differences of the sounds of different instruments and therefore affects (often positively) the mix.

REDUCING NOISE AND LEAKAGE

Noise and leakage are usually controlled by the gate and the expander. In general you are trying to set the threshold above the level of the noise and leakage but below the level of the sound that you want to keep. The difficulty in setup of these devices is that there often is not a clear level that is always above the leakage and always below the level of useable sound that you want to keep. The result can be the notes of the sounds you want to keep being chopped off at the beginning or end.

Usually the release time can be adjusted upward to allow the sustain of the guitar (or the ring-out of the cymbal) to finish before the gate turns down the level. Notes getting chopped off at the beginning is a function of the attack time parameter and more difficult to correct. Two common things that can be done are:

Reducing Expansion Ratio - Expanders generally have a ratio of 1:2 and gates have a ratio of 1:20 or higher. Using the expander ratios tends to make the beginnings of the notes less affected by the dynamics processing.

Reducing Range - A reduced range means that the unit will turn down the softer (noise and leakage) signals less. If you are having difficulty with the attack of the notes, reducing the range to 6-10 dB keeps a fair amount of noise and leakage reduction but makes the attack on the notes more natural.

PREVENTING OVERLOAD AND PROTECTION

The overload of a digital recorder sounds really bad. The clipping of a power amplifier, likewise, sounds bad. Unexpected loud transient signals (like someone dropping a microphone) can cause damage to speakers and amplifiers. A limiter with a very fast attack time (0 ms if possible) can prevent any peak that would cause bad-sounding overload or even damage. Generally peaking the threshold is set just below the level of clipping (overload) and the release time is also set fast (5-10ms) and limiters used for protection use hard knees.

INCREASING SUSTAIN

Increasing sustain of an instrument uses settings that normally would cause breathing. The long release time allows the instrument's "ring out" to be turned up, allowing you to hear the sustain for longer. Sustain is increased with fast attack and slow (400 ms or more) release time. Compression ratios of 4:1 to 8:1 are often used with a soft knee.

DEESSING AND OTHER USES

Using the compressor to remove unwanted blasts (such as accented "S" sounds and pops) and advanced dynamics processing techniques involve the use of equalization and the sidechain input of the dynamics processor. These are discussed in a future chapter entitled Advanced Dynamics Signal Processing.