•   Wwise Compressor
•   Wwise Expander
•   Wwise Peak Limiter                    

This document provides a quick overview of the behavior of these plug-ins, the signification of their parameters, general remarks, etc.                

Compressor/limiter

Overview                 

The objective of dynamic compression is to reduce the dynamic range of a signal (often with the objective of applying make-up gain in order to boost the signal). Limiting is in essence compressing a signal with more "extreme" parameter values.                
   
The principle behind the plug-in algorithm is to detect the RMS power (i.e. not its maximum amplitude) present in the input signal and apply adaptive gain control. When the signal is soft (below the threshold parameter) no gain reduction is applied, when the signal gets louder and exceeds the threshold, the compressor kicks in and starts reducing the gain. The speed at which gain reduction is applied is determined by the attack time parameter. Similarly, the speed at which the compressor kicks out to no gain reduction is determined by the release time. The ratio factor determines how severe dynamic compression will be when it exceeds the threshold. For example, a ratio of 4 means that for every 4 dB over the threshold, the output will only get 1 dB louder than the threshold.

Parameters

•   Threshold (dB): Value in dB above which the compressor starts to operate
•   Ratio (no units): The slope of the compression above the threshold. For every dB that exceed the threshold in the input, 1/ratio dB over the threshold will be observed in the output when the compressor reaches its steady state. 
•   Attack time (sec): Time taken for the compressor to reach 90 % of its gain reduction target value. 
•   Release time (sec): Time taken for the compressor to reach 90 % of its ramp towards no gain reduction. 
•   Output gain (dB): Gain applied after the compression algorithm to make-up for potential gain losses. 
•   Process LFE: Will apply compression algorithm to LFE channel (when present) if engaged, otherwise the LFE signal is passed through unaffected.
•   Channel link: Gain reduction will be the same for all channels if engaged. Power estimation uses all incoming channels. When disengaged, effectively the same as parallel processing of all channels.  

Expander/noise gate

Overview

An expander is exactly the opposite of a compressor: it applies no gain reduction above the threshold but further reduces the amplitude of softer parts of the sound. A noise gate, is simply an expander with extreme settings, using an expansion ratio so high that literally everything below the threshold is completely blocked. The algorithm and parameters are all the same but interpreted differently.                
   
The plug-in algorithm detects the RMS power (i.e. not its maximum amplitude) present in the input signal and apply adaptive gain control. When the signal is loud (above the threshold parameter) no gain reduction is applied, when the signal gets softer than the threshold, the expander starts further reducing the gain. The speed at which gain reduction is applied is determined by the release time parameter. Similarly, the speed at which the expander returns to no gain reduction is determined by the attack time. The ratio factor determines how severe dynamic expansion will be when the input is below the threshold. For example, a ratio of 4 means that for every 1 dB under the threshold, the output will get 4 dB softer than the threshold.                

Parameters              

•   Threshold (dB): Value in dB below which the expander starts to operate  
•   Ratio (no units): The slope of the expansion below the threshold. For every dB that under the threshold in the input, ratio dB under the threshold will be observed in the output when the expander reaches its steady state.  
•   Attack time (sec): Time taken for the expander to reach 90 % of its ramp towards no gain reduction.  
•   Release time (sec): Time taken for the expander to reach 90 % of its gain reduction target value.  
•   Output gain (dB): Gain applied after the expansion algorithm to make-up for potential gain losses.  
•   Process LFE: Will apply compression algorithm to the LFE channel (when present) if engaged, , otherwise the LFE signal is passed through unaffected.
•   Channel link: Gain reduction will be the same for all channels if engaged. Power estimation uses all incoming channels. When disengaged, effectively the same as parallel processing of all channels.  

Peak limiter

Overview 

A peak limiter is usually applied at at the top of the audio pipeline hierarchy (the master bus) to prevent any distortion due to clipping in the output signal. While the Wwise Compressor plug-in algorithm may be used as a limiter by setting the ratio to a higher value (e.g. >= 10 ) and using a shorter attack time and longer release time, the averaging of power implicit to RMS based detection methods does not guarantee that the signal will not exceed the threshold by large values for transient signals. To prevent this condition, a look-ahead time is allowed on the peak limiter to ensure that the attack time will have reached its target value before hitting the offending transient peak. This measure will introduce a latency of the output signal determined by the look-ahead parameter. Otherwise behaves similarly as the Wwise Compressor, only that for the time window specified by look-ahead time, only the peak value will be considered (no RMS averaging).                

Parameters

•   Threshold (dB): Value in dB above which the limiter starts to operate  
•   Ratio (no units): The slope of the compression above the threshold. For every dB that exceed the threshold in the input, 1/ratio dB over the threshold will be observed in the output when the limiter reaches its steady state. For reasonable look-ahead time parameter values, this condition will always be fulfilled.  
•   Look-ahead time (sec): Time offset at which signal peak detection is operated. Also determines directly the latency introduced by the Peak Limiter.  
•   Release time (sec): Time taken for the compressor to reach 90 % of its ramp towards no gain reduction.  
•   Output gain (dB): Gain applied after the limiter algorithm to make-up for potential gain losses.  
•   Process LFE: Will apply the peak limiting algorithm to the LFE channel (when present) if engaged, otherwise will keep sync between LFE and other channels based on the look-ahead time without peak limiting its signal.
•   Channel link: Gain reduction will be the same for all channels if engaged. Power estimation uses all incoming channels. When disengaged, effectively the same as parallel processing of all channels.  

  General remarks

• There is no interpolation of ratio control (RTPC parameter) for all 3 plug-in (for performance reasons). Changing this parameter during playback may lead to signal discontinuities on very clean signals. If this is reported to be a problem for normal (non-calibrated) signals we can put it in.
• Sounds with DC offset may alter results of compression/expansion because the sidechain detection algorithm will be offset considerably. DC offset should be removed prior to use in Wwise.  
•  Compressor/Expander and Peak limiter are non-linear audio processes. What this really means now is that order of processing is highly relevant (i.e. results will be different if you apply gain before or after the effect ).  
•   For the first audio buffers, the processing algorithms are said to be in non-steady state. Because the estimation of signal power operated by the side chain has no knowledge of past, its estimation of signal power may be erroneous for a small period of time (only if the compressor is already engaged at the start of the sound). A solution to this (if it's a problem at all) is to start compressor with a ratio of 1 and shortly after bring it to the desired value.  
•   If the compressor applies some gain reduction and is suddenly bypassed, a signal discontinuity will be heard. A  potential solution is again to gradually bring the ratio to 1 prior to bypassing the effect.