Dynamics Function

The Dynamics section comprises of four separate applications, Compressor, Limiter, Gate and Expander. Although these applications contain several control types that have common functions, the operation, ranges and laws of these controls have been optimised carefully for maximum flexibility within the intended specific use of that section. The very wide control ranges offered within the applications are accommodated in the GUI using specific control laws that encourage experimentation over a very wide parameter set without the loss of finer control sensitivity and detail. Careful consideration has also been given to the dynamic behaviour of the time constants, as this factor is largely responsible for the sonic character of the dynamics application.

In order to make best use of this application it is necessary to gain a basic understanding for dynamics processing in general and the particular architecture of the Sonnox Oxford Dynamics plug-in. The following sections address these issues.

Basic Compressor Architectures

There are two basic types of dynamics architecture in common use; these are often termed as feed-forward and feedback types. The feedback type uses it's own output to compute required gain reduction. This method had an advantage in early analogue compressors because the complex and largely unpredictable laws of early gain reduction elements could be somewhat decoupled from the total level transfer characteristic of the application (because the design made use of level feedback). With the introduction of better solid state VCA's and accurate logarithmic side chain processes, this method has largely been abandoned in favour of the feed forward model since it has a much greater degree of control parameter separation and intrinsic accuracy. The Sonnox Oxford Dynamics section is a feed forward type processor. By that we mean that the gain controlling side chain element of the processor works by evaluating the programme level at the input, calculating the required output gain by dead reckoning. This widely conforms to the architectures of most popular modern analogue dynamics sections employing voltage-controlled amplifiers. In a digital design the feed-forward model has additional advantages, which include the possibility of extremely well controlled and variable time constant laws and sonically accurate gain control elements. Also look-ahead processing (using delay) allows gain control to be initiated in advance of the signal without signal quality loss.

Back to Top

Compressor Level Control Functions

There are two main factors that describe compression function; level versus gain function, which is generally assumed to be independent of the time constants, and dynamic gain function which exhibits more complex dynamic behaviour over time. In explaining the operation of the compressor section, it is useful to split these categories. The following section refers to the level versus gain behaviour of the compressor application of the Sonnox Oxford Plug-in.

The Sonnox Oxford plug-in employs logarithmic side chain processing, which means that all signal parameter setting (and time constant action) occurs in the ‘decibel' domain. This makes it possible for all control functions to remain independent and therefore provides the greatest level of control for the user. To get the best results from the application it is useful to gain an understanding for the specific effects of the control parameters.

Compressor Threshold Control

The threshold control sets the level (ref dBFS) that compression and gain reduction will begin. The control has a linear decibel law over the range. The following is an illustration of the threshold control operated at 5dB increments with the ratio at max (1000:1) to –20dBr.

Compressor Ratio Control

The ratio control sets the rate at which gain reduction will occur when input level goes beyond the set threshold level. The control has a 1/Ratio law so that gentle compression can be achieved despite the wide range of the control. I.e. from 1:1 to 2:1 ratio occurs over the first 50% of the control range and 4:1 ratio occurs at 75% control rotation etc. Full limiting is achieved at 100% control rotation.

Compressor Soft Ratio Control

The soft ratio function provides a gentle, minimum rate transition between the region below the threshold and the compressed region of the curve. A further threshold below the main threshold control setting, defines the start of the soft curve. The program signal is therefore compressed progressively harder as it gets louder within this region until the full compression defined by the ratio control is achieved. The adjustment range via the Soft button is from 0dBr to –20dBr in 5dB increments.

Compressor Gain Control

The gain control allows manual compensation for level loss during compression to a maximum of +24dBr. This control operates independently from all other settings and can be considered as an output level control. The gain control is applied to all Dynamics functions but operates only when the compressor in ‘IN'.

Using Level Control Functions

Since all level control functions in the Sonnox Oxford Dynamics Compressor operate entirely separately, a very high degree of control for a wide range of common use is possible, in particular the plug-in does not impose any particular style constraint on the user. The section explains some of the commonly used techniques and how they can be achieved using the Dynamics plug-in.

In the most general terms the extremes of compression usage falls into two main categories, dynamic level control and sound effect generation. For simple level control, such as controlling performance variation in vocals, instruments and final programme, we most often require the most transparent compression with minimum artefacts due to the dynamic control. However, to generate sound effects and distortion the reverse is true and we need to make the sound character of the compression a dominant part of the final result.

In order to understand how we achieve these two different styles of result we must remember that in general we are much more sensitive to the rate of change of level than we are relative gain. So in order to generate prominent sound effects from the compressor we need to generate significant rate of change of gain action by using both hard compression gain curves and the creative use of time constants. However for general unobtrusive level control we should be avoiding all of this and opting for the gentlest compression gain curves and least obtrusive time constant settings that integrate well with the programme style.

General Programme Compression

There are 2 main basic philosophies that underlie approaches to unobtrusive compression. This section aims to explain these concepts, make comparisons between them and show how enhanced results can be achieved using the Sonnox Oxford Dynamics plug-in compressor.

‘Least Possible' Approach

The first and most obvious, which I will call the ‘ least possible approach' is to leave the majority of the programme uncompressed, forcing the compression to deal only with the louder passages. This method has a definite psychological advantage in that one gets the feeling that the majority of the programme remains unaffected. There is also some possible technical merit (especially for legacy designs) in that the compressor is working ‘less often' and over a restricted range, thereby avoiding some of the potential errors in the application. The above plot illustrates this kind of approach. The idea is that the programme goes ‘over' on the loud passages so we seek to control the loud portion only by setting the threshold relatively high (-5dBr) and setting the ratio high enough (i.e. between 3:1 & 4:1) to prevent the over. With the Sonnox Oxford compressor the threshold and gain make up controls can be used to accommodate this approach over a 24dB range of relative input levels without change in the sonic character of the programme.

This method has the major disadvantage of risking increased rate of change disturbance because the transition between non-compressed and compressed programme regions is sharp. Therefore there is a considerable reliance on longer time constant settings in order to reduce the sonic effects of the compression. I.e. we need to seek to control the rate of change using time slewing rather than level progression. Whilst this approach would naturally form a good basis for using the compressor as a sound effect, it is less suitable for composite programme control.
One way to alleviate the compression transition effect is to smooth it out using the soft ratio function. The above plot shows the action of soft ratio set to 5dB applied to the previous settings. Starting the compression earlier and increasing the compression ratio up to the max level point smoothes out the transition point at the onset of compression. The main advantage of this method (often referred to as ‘over easy') is that faster time constants can be used before the compression artefacts become too obtrusive. This means that the programme can be made to sound louder and more present without increasing peak levels, despite the fact that more of it is being compressed. Also, the use of faster time constants further reduces peak overshoot, so there is less need to employ limiting to the signal chain.

It should be noted that with the Sonnox Oxford plug-in, the application of the soft function will always result in the same maximum level output. This means that you can apply the soft ratio function at will, with only a minimal need to adjust the threshold and ratio controls.

Combined Approach This concept advocates that a more transparent sounding compression can be achieved if a relatively large portion of the programme level range is under continuous compression. The rationale here is that the rate of change disturbances are minimised because the compressor spends less time going over the onset of compression transition range. But the downside is that peak loudness is less well controlled, therefore quite heavy additional peak limiting is sometimes required to tailor the performance into an overall mix. This plot shows one method to achieve maximum loudness and presence for programme where dynamic range is unwanted (i.e. Pop music, spoken commentary etc). The intention is to get the prominent parts of the programme as loud as possible without incurring too many compression artefacts that would require long time constants to fix. The gain is set to maximum (+24dBr), the ratio is set to 1000: 1 (limiting) and soft ratio is selected at 15dB, then the threshold is reduced (to around –25dBr) until max level is achieved. The action is to compress the programme to ever-greater degrees as it gets louder, until it is fully limited at maximum output. The selection of the 15dB soft ratio curve is a compromise between unwanted compression artefacts, available gain and total loudness. It should be noted that as the top 10dB of programme has virtually no dynamic information at all so this sort of approach is likely to gain favour within the current Popular music idiom. It can also be very useful in vocal situations where a fast and smooth action is needed with minimal distortion. This plot shows a minimum intrusion type curve using a ratio of around 2:1 and a soft setting of 10dB. This style of compression will bring up the softer passages by around 12dB and progressively compress the loud passages into half the original dynamic range in a very subtle and unobtrusive fashion. Settings of this kind can tolerate fast time constant settings without obvious sonic artefact and are therefore very effective in classical music programme or other situations where a good degree of dynamic realism needs to be preserved.

Back to Top

Timing Functions

The relationship between gain compression and temporal behaviour is absolutely crucial to successful dynamic control, whether the user is aiming for unobtrusive control of levels or a completely stylised sound effect. Much of the artistic characters of the massive range of renowned compressors are determined by subtleties in the timing behaviour of these units under complex musical signal conditions.

In order to encompass as much of this valuable artistic legacy as possible, the Sonnox Oxford compressor plug-in incorporates a very wide range of timing control, using both linear dB and exponential dB time dependency curves. In this way the plug-in can be used to create the widest range of artistic effect without imposing an overwhelming and unavoidable character of it's own. The plug-in should therefore be considered as a powerful and flexible tool rather than a specific ‘style' of compressor. The following section describes the operation of the compressor timing controls and time dependency behaviour.

To generate the timing illustration plots the following stimulus was used for reference. This is a tone burst signal that consists of 0.2S of full level signal preceded and followed by 0.8S of signal at –20dBr.

Compressor Attack Control

The attack control varies the timing behaviour of the compressor at the onset of gain reduction, either due to the arrival of a peak level above the threshold or the further increase in a level already above the threshold.
The three plots above show the action of the attack control from minimum through mid range to maximum setting.

Compressor Release Control

The release control varies the timing behaviour of the compressor during the recovery period after a gain reduction, when the signal level has reduced from previous levels above the threshold. The above plots show that action of the release control from minimum to half range (maximum range is too long to show successfully on this plot).

Compressor Hold Control

The hold control varies the amount of time between a reduction of levels above the threshold and the onset of the release time. It provides a period after gain reduction where a slower rate of recovery occurs. The above three plots show the action of the hold control from nearly minimum to 75% of maximum with an intermediate position with the release set to minimum. When the compressor is in recovery after a loud passage, the full value of the release setting time is reached only gradually over the period of the hold time. This produces and extended period when the release time is maintained at slower rates in order to allow faster gain recovery without excessive distortion at low signal frequencies. Since the release time is effectively multiplied by the hold time, very long total gain recovery times can also be obtained by using the release control in conjunction with the hold control.

Compressor Timing Laws

The Sonnox Oxford compressor offers three compression types called Normal , Classic and Linear. The time constant laws employed define the major difference between these types.

The Sonnox Oxford Dynamics uses both exponential dB/time and linear dB/time laws as illustrated below.
The Sonnox Oxford dynamics time settings are displayed as real time constants when in exponential mode. In linear mode the display is scaled to correspond to the time required for a 10dB change of level. This provides a degree of parity between the perceived timings of the two laws, which facilitates comparison and selection between compression types. These two laws have very different sonic characteristics.

Exponential dB Timing (Normal & Classic types)

The exponential/dB curve is by far the most popular law used in a great many well-respected compressors and is the natural result of more recent analogue units employing logarithmic side chains and resistor/capacitor time constants. The exponential/dB law has some interesting characteristics. Firstly, the time taken to complete a compression event tends to stay the same however large the dynamic signal excursion is. Also since the peak rate of change of gain increases with dynamic excursion, the resulting harmonic content due to compression tends to follow the loudness of the programme in a way the ear expects. This helps to mask the effects of the compression and thus provides the most forgiving solution, being tolerant to differing timing settings and programme material. This makes it the best choice for general compression use and overall dynamic control of complex musical programme. The plots above illustrate the action of the exponential law, the first shows 10dB gain reduction (scaled for comparison) and the second one shows 30dB. As can be seen, the initial rate of change is much increased in the attack period and the total time for attack is similar despite the increased level transition.

The Classic type selection is a subset of the Normal type with timing controls fixed to nominal values to match a range of popular legacy units. All other controls behave as the Normal type. This type selection is quick to set up and is most useful as a general-purpose channel compressor.

Linear/dB Timing (Linear type)

The linear /dB law in some respects has the reverse behaviour of the exponential law. Because the rate of change of gain is constant (as set by the timing controls), the greater the signal dynamic excursion the longer the compressor will take to complete a gain change. Also, since the total time that the compressor spends in attack or decay is proportional to the size of the gain excursion, the harmonic content of the compression artefacts will seem to reduce in frequency content the louder the signal excursion is. This type of compressor is useful for generating dynamic sound effects because the sonic character of the compression is much more affected by time control settings and programme material than the exponential type. The above plots illustrate the action of the linear law. The first shows 10dB gain reduction (scaled for comparison) and the second one shows 30dB. The rate of change is the same during the attack period and therefore the total time for attack (and subsequent release) is increased with greater level transition.

Please note that the linear timing law is generally unsuitable for the control of programme dynamics and modulation levels because of its unnatural sounding relationship between level excursions and perceived overtone generation.

Using Compression Timing Functions

The setting of timing functions can drastically affect the sonic character of compression and there are many different approaches to compression timing, often in pursuit of ever-changing fashion. Therefore there is no right or wrong approach to this task. However to successfully build up your own portfolio of artistic sounds using a variable parameter compressor such as the Sonnox Oxford plug-in a basic grounding in the sonic effects produced by timing is useful. The following section sets out to describe the basics of timing settings and the range of sonic effects available from the Sonnox Oxford plug-in. By changing time constants many effects can be generated such as:



There are many approaches to programme level control, which are largely decided by whether one is trying to get maximum loudness and excitement, enhancing reproducibility at low reproduction levels or just trying to control overloads. In mastering situations either or all of these may be appropriate along with many other subtleties such as matching impressions between tracks destined for an album release. The following is a general description of the some effective approaches and starting points.

Fast as Possible Approach

To obtain absolute maximum modulation and minimum dynamic range the best approach is to set release times to minimum, increase the hold time just enough to the reduce LF distortion to acceptable levels and increase the attack time just enough to allow some overshoot on percussive peaks, in order to retain some impression of programme dynamics. The appropriate level of compression can then be obtained using the threshold, ratio and soft ratio controls. The overshoots produced by the attack times can be controlled by the use of the programme limiter section.

Natural Dynamics Approach

To obtain a more natural compression a good starting point is to set the attack and release controls to mid positions with hold control at minimum (this is the fixed setting of the Classic compressor style). This approach aims to match to some degree the dynamics of the ear's response and recovery from loud sounds at relatively high sound pressure levels. Variations on these moderate settings can yield realistic results if appropriately adjusted to suit the intended reproduction levels of the programme.

Slow and Gentle Approach

For level control with the least possible intrusion the method is to set the attack and release times to the longest possible times, perhaps with the addition of the hold control to increase the release times further. This ensures that the highest levels within the programme are controlled gently and the gain recovery in the quiet passages is almost imperceptibly slow. This method is most effective when used in conjunction with the larger soft ratio settings, as this ensures that compression commences well before the target maximum level, providing a degree of prediction.

Artistic Effects

The manipulation of timing within compression can create some very useful sound effects. In particular gain overshoots produced by slow to moderate attack times can be very useful at tightening up soft percussion sounds. However a note of caution is needed for users of workstation applications in that effects such as these may cause unexpected programme clipping that may prevent the available range of possible sounds being fully appreciated. In particular, for systems that lack overload margin between plug-ins or within their mixing structures, the extra short term peaks produced by creative compression may be prematurely clipped within the host application because essentially there is no range (headroom) to accommodate them. In this case a reduction of input levels and/or a suitable reduction of the threshold and gain make-up values may be needed to fully realise the new sound within the intended mix.

Back to Top

Limiter Functions

From a level profile perspective the limiter function is essentially the same as a compressor set at infinite ratio. However the use of much faster attack times with a somewhat different dynamic behaviour allows faster and more effective reduction of peak levels. Historically limiters were developed mainly for radio transmission systems where absolute limits on modulation were needed.

In this case simple saturation was not useful since the HF energy produced by signal clipping could still breach modulation limits. The earliest limiter designs were mostly fast attack slow release types to minimise audibility problems. Since then many different designs and much more complex methods have been developed. And more recently limiting and its side effects has almost become an artistic tool in itself, being partly responsible for the recent trends for absolute maximum modulation and loudness, currently favoured amongst producers and broadcasters of popular music. As a result of this expanding trend for stylised dynamic control, the demarcation between limiting, compression and even EQ has become increasingly blurred as an increasing number of ever more complex devices become available that make use of all of these functions in a quest to produce impressive results.

The Sonnox Oxford plug-in limiter however is presented as an entirely separate and direct process that is designed and optimised specifically for highly efficient and musical peak limiting functionality. Despite being simple and intuitive to understand and operate, superior results are obtained from highly accurate level and timing behaviour making optimal use of look ahead processing that acts on signal peaks prior to their arrival at the gain control element. Although the limiter is designed to complement the compression section of the Plug-in, it can be equally well be used as a stand alone application for enhanced peak level control and programme modulation maximisation etc.

General Description

The limiter function controls are presented similarly to the compression function except that there is no ratio control. The time constant functions, although similar to the compressor, have different ranges. In particular the attack time can be adjusted to be much faster than the compressor. It should also be noted that the threshold levels are related to the output level of the whole dynamics application (rather than the input level), such that any gain increases produced in the compression section and gain make-up do not affect the calibration of the limiting target levels.

General Limiter Operations

There are many methods and approaches to programme limiting that are favoured amongst users. In general however these fall into two main categories as described below.

The least intrusive kind of limiting is achieved with a fast attack and a relatively slow release, adjusted to suit the general timing of the music. Very long decays, often favoured in classical music productions, can be achieved with the Sonnox Oxford limiter by using a combination of both release and hold times. Due to the look ahead processing, the Sonnox Oxford limiter also has a significant range of attack time that can be used without peak level overshoot. This means that a slower and gentler limiting can be achieved during the onset of loud passages within the programme material without breaching maximum levels as set by the threshold control.

Maximising the Loudness Using the Limiter

A recent popular use of programme limiting aims to maximise the relative loudness and average modulation of the music by reducing the short-term peak levels within the programme waveform. This makes it possible to increase the overall volume (gain) of the piece without getting obvious overloads (red lights) on the final mastered work. The success of this method depends on the amount of peak reduction can be obtained without objectionable loss of quality to the programme. Maintaining the quality of the peaks without reproducing their maximum levels relies on a degree of ‘peak remodelling'.

The basic approach to this function is to initially set attack, hold and release controls to minimum for fastest possible action. Then reduce the threshold control to progressively reduce the gain of the programme, only during the very short periods where peaks occur. The sound of the peaks can then be adjusted by increasing the attack period to soften the edges such that a degree of realism is retained despite the reduction of instantaneous levels during that period. Since the Sonnox Oxford limiter has a fairly large look-ahead period, it is possible to increase the attack time significantly without allowing the peak levels to pass. It is therefore possible to re-model the limited peaks considerably in real time, retaining good programme realism despite large degrees of peak reduction. If it is necessary to increase recovery times after peaks, for instance due to LF modulation content, it is better to use the Hold control for this purpose, as the recovery time is faster for a given overall period than by increasing the release time. The programme level can then be increased so that the new peaks just reach maximum level again. Depending on the material, average programme modulation may be increased by 6dBr or more by this method.

Back to Top

Expander Functions

The Dynamics expander section control functions are presented similarly to the Compressor and Limiter sections, except that a Range control is added. The architecture of the expander conforms to what is often described as ‘downward expansion', which means that the application only works to attenuate existing signals below a set threshold and cannot produce any additional gain for signals above the set threshold levels. In practice this means that, although the purpose of the expander is to increase dynamic range, it can only achieve this by reducing the signal level in the first place.  The Threshold control sets the level beneath which the expander will become active.

The Ratio control sets the gain slope rate that signals below the threshold will be attenuated with respect to the input levels.
The Range control sets the bottom limit that any attenuation caused by expansion can reach.

Expanders are used for many purposes, both technically and artistically. These include background noise reduction and file clean up, ambient noise disturbance reduction in live recordings, presence and dynamic profiling of instruments and creative control of ambient reverb etc. Because the Sonnox Oxford expander has 20 samples of signal pre-view, it is particularly useful for instrument dynamic profiling as it can act on the signal before gain control actually occurs, meaning that one can actually profile the leading edge of a percussive attack for instance.

Back to Top

Gate Functions

The gate section controls of the plug-in are presented in the same way as the expander function except that there is no ratio control. Programme gating has become ever more popular in recent years since its inclusion in some professional console systems. Originally intended as a technical tool for tape noise suppression and such, many useful and often fashion changing artistic effects have been achieved using gates. The Sonnox Oxford plug-in gate has many advanced design subtleties that are based on long experience in the artistic use of gating.

The basic architecture of a gate is similar to expansion with infinite ratio and therefore acts much like a programme switch. Signals below the set threshold are cut and only those above this are passed to the output. The gate also has level hysteresis such that once opened by signal above the threshold, a subsequent signal level reduction of 4dBr is required to close it again.

The residual signal level for programme below the threshold is set using the Range control, as in the expander section. Since the gate effectively operates on signals before gain control is affected, the Attack control can provide a significant range of attack profiling without missing the peaks within the programme. This provides the capability to change the sonic character of the attack period of the gate.

The Hold function provides time hysteresis by effectively providing a delay after gate opening before subsequent closure can occur, which is very useful for trimming gate activity to match programme event timing durations and nested rhythms within the music.

The Release control provides the ability to tailor the recovery period to match the programme material and provide artistic effects.

Back to Top

Side Chain EQ Functions

A high specification two-band side chain EQ is provided to allow de-essing and other modifications of side chain frequency response. The EQ processing section may be routed to either the dynamics side chain or the main signal path or both simultaneously, via independent EQ-Side chain and EQ-Signal selector buttons.

Back to Top

Harmonic Generation

Although there is much discussion of the importance of level and time responses in defining the character of dynamics and compression applications, these are not the only parameters that affect the user's experience of these applications. As stated within the previous texts, many original and much loved legacy units have gained their reputations for sound quality based on inherent characteristics produced as by-products of their signal chain and gain control regimes. Indeed some tube-based units are favoured as much for their basic sonic character when inserted in the signal path, as for their programme compression characteristics. This is almost invariably due to accumulated signal path distortion and novel overload characteristics when in static mode, or a more complex combinational modification of these effects during dynamic gain control conditions.

In recognition of the important artistic dimension that these effects confer to the dynamics process, the Sonnox Oxford Dynamics plug-in includes processing functionality destined to endow an appropriate harmonic character to the dynamics function, in a similar vein to those often experienced when using legacy units of renown. However due to the controllability and accuracy of the digital signal path, this character can be controlled by the user, rather than simply being an inherent and unavoidable side effect of the application.

Warmth Function

This process confers additional loudness, punch and definition to the sound of the dynamics section. The operation of this process is to impose a harmonic profile onto the signal that increases the density of higher value samples within the programme, in order to boost average modulation levels without an increase in peak levels or the risk of digital clipping. The warmth function is engineered to achieve this without loss of dynamic information within the programme.

Another very important purpose of the Warmth function is to accommodate internally generated levels up to 6dB greater than notional digital maximum without causing increased peak levels or hard clipping. The main advantage is that short term level peaks and overshoots, resulting naturally from the artistic use of compression, can still provide harmonic information to the programme even though they may represent levels above maximum digital modulation, that would be subjected to hard clipping if left untreated.

The process also adds subtle warmth to the programme material that is reminiscent of tube systems and is similarly tolerant of overloaded or previously clipped signals, avoiding much of the harshness associated with these conditions. Significant artistic effects may therefore be achieved by deliberately overdriving the warmth processing by increasing the compressor gain make up beyond normal levels and allowing the warmth processing to control the peak signal levels.

When selected, the Warmth function operates on the complete signal chain of the dynamics plug-in and is therefore applied to the signal even if no compression functions are operative.

Back to Top