Dynamics Processing Explained via PT DYN3 plug-ins

by eddie ciletti
3rd March 2014

The difference between a compressor and a limiter - you might even say the 'magic' - is really a matter of side chain settings - Attack, Release, Ratio, Knee and Threshold.  This article, combined with the images below, will explain examples of Peak Limiting (left) and Compression (right).  The difference between ProTools DYN3 dynamics plug and other dynamics plugs SHOULD be the emulation of the various hardware non-linearities (distortions) that come from audio transformers, signal and side-chain amplifiers and even 'mis-aligned' or nearly broken signal processors.  In Art, all of the possibilites are acceptable if the means justifies the end result.

For Peak Limiting, note that the Attack and Release are at their fastest  - 10-micro-seconds (uS) and 5-milli-Seconds (mS).  The 'Knee' is set to "0," the hardest setting and the ratio is set to 100:1, the highest setting.  In this case, all that is necessary is to adjust Threshold (lower the ceiling) until theamount of gain reduction (GR) is between 3dB and 6dB.   On transient-rich material like drums, 3dB should be almost undetectable to the ear, 4.5dB GR is still very acceptable and anything more than 6dB will be obvious.  Sometimes, you don't want obvious. 

In this example, the snare consists of the 'stick' (attack) and its resonance, the 'tone.'  For the moment, all we want to do is reduce the peaks created by the stick and not disturb the tone, because peaks are what get in the way of making the signal louder.  A fast release ensures that the tone will not be affected.  To assist in visualizing, the raw signal is at top and the processed signal (via AudioSuite) is at the bottom.  Using Audio Suite is a good way to see if the processing is serving its intended purpose.  Now, some people will say 'just listen,' but again, sometimes you want 'surgical.'

All dynamics processors implement Threshold in either of two ways.  The DYN3 Threshold merely lowers the ceiling  until the signal 'crashes' into it.  On other limiters, Threshold increases gain until the signal hits the ceiling.  I prefer no gain change so I can have control.  I prefer to keep levels conservativwe withint protools because I feel the metering is not particularlay accurate - PT is so busy with graphics that it doesn't have time to report all of the 'overs.'  Note that an 'over' is not one instance of clipping, but several instances in close succession.   Some people use external and internal clipping on purpose as their 'peak limiting.'   Since audio recording and mixing is both art and science, 'it's all good,' except  you should know that clipped transients on stereo tracks, stereo sub mix or stereo master can degrade the depth of the stereo image.  I prefer NO overs.  I avoid hot levels going into some plugs - in digital, headroom is still important - and I avoid make-up gain unless necessary.  If i see a 'stuck' RED OVER, I figure out why and fix it.

Just as some peak limiters add gain when Threshold is adjusted, I prefer those that don't and I rarely use make-up gain unless the processed signal is too quiet.  Compression essenttially raises low level signals and reduces louder signals.  The signal below right is a bass.  You can see the 'spikes' of the string being plucked and you can see the notes.  After processing, the spikes are controlled and the notes are more consistent.

For Compression, the Attack is much slower than peak limiting - 8-milliSeconds (mS) to 12-mS Attack is fast enough to preserve some of the punch.  Release speeds can range from Fast (125mS to 250mS), Medium (250 to 750mS) and Slow 750mS to as many seconds as the device will allow.  In this case, the Knee is at max soft (30), the ratio is low (3.2:1) the Attack is 16.7-mS and the release is 575.7-mS.  Threshold can be adjusted for more aggressive amounts of Gain reduction (GR), although like Peak Limiting, I like to 'limit' processing to 6dB primarialy because I often have multyiple sigjnal processors in a mix.  For example, I may use just a touch of compresson on the bass track, then submix bass and kick together with another touch of compression to glue them together, then submix the kit and use a dyn3 in series with a McDSD ML-4000 multiband compressor (along with its stereo peak limiter).  All together that coudl be 12 dB of gain reduction!

Keep in mind that raising the signal level 6dB doubles the signal voltage.  Reducing the signal level by 6dB - through gain reduction or otherwise - reduces the signal voltage by half. 

IMy multi-band processor of choice is the McDSP ML-4000, I prefer it over other, more popular multi-band plugs.  Here is a sample ML-4000 tweak for killing snare ring.   I will be posting an updated article about my unconventional tips and settings in the near future.  If you are impatient, email me!