Audacity 101 – Compression

Welcome to the third instalment in our series on the open source audio editor Audacity.
Although mainly aimed at beginners, This piece by Robert Carlberg is so interesting that we are making it this week’s instalment . Thank you Robert.

Closely allied with Normalisation is the process of Compression.  Early studio technology for compressing signals gave the process a bad name, as they were sudden and audible.  They caused notable “clamping” in loud sounds, such as a bass drum, which was also known as “audible gating.”  When the incoming signal exceeded a threshold set by the engineer, this caused the compressor to “clamp down” or “gate” the signal that followed, giving an unnatural release curve after loud sounds.  This “gated bass drum” sound is nearly universal in any studio recording, particularly in LA studios, made between 1970 and the early 1990s.  It is still commonly heard in many forms of music, having become so accepted that many people have come to expect it.  People who have never heard live drums actually think they should sound like that.  (Drummers sometimes stuff their bass drums with pillows or blankets to get “that sound.”)

It wasn’t until the introduction of the Compact Disc in 1982, and their widespread adoption later in the decade, that compression algorithms became more sophisticated.  Compact Discs have ten times the dynamic range capability of LPs, cassettes and radio used in the 1970s.
Compressors were slow to evolve however, as the sound of gated audio was by then readily accepted.  Breakthrough albums like “Tricycle” by Flim and the BBs (1984), “90125” by Yes (1983) and “Amarok” by Mike Oldfield (1990) showed what an uncompressed signal could sound like.
Unfortunately, many engineers decided compressors should be used to bring up ALL signals, even the quiet ones, to the limit of clipping.  This came to be known as “the loudness wars.”  When comparing two clips of music, listeners tend to prefer the louder one, even if it’s not the most realistic.  Compact Discs allow for 0dB (full scale) signal without clipping, and engineers took this as a green light to push everything to the limit.  To a very large degree, this is what gave CDs a bad reputation in the ’90s among many listeners, as rampant “brickwall compression” by engineers made everything sound like AM radio again.
But modern compression algorithms are sophisticated.  They have something called “lookahead,” where they begin their clamping well before the peak, giving a smooth, natural-sounding reduction in the overall volume.  Couple this with subtle, non-obvious raising of the volume between the peaks, and compression can improve the listenability of most types of music.
A quick listen to Flim and the BBs or “Amarok” will convince you — uncompressed music is simply scary.
An example of a usable compression setting in Audacity is depicted below.

In this example, we have a one millisecond lookahead — fast enough for percussion transients — and a two millisecond release, short enough that the percussive “hit” still sounds like an impact.

The Threshold setting, where the compressor begins compressing, is set to -3.0dB, well below the maximum signal strength of 0dB.  Incidentally, decibels are logarithmic, meaning a three decibel change is a halving (-3dB) or doubling (+3dB) of the perceived volume.  Therefore -3.0dB is HALF the maximum volume.  But the setting after that, the “Make-up gain,” is also set to 3.0dB, in this case PLUS 3.0dB. That means the peak still peaks at 0dB, it just reaches that peak in 3 milliseconds (including attack and release) instead of being instantaneous. The peak is not squared off — as in clipping — but rises and falls naturally and smoothly.

The Make-up gain setting also means that all signals BELOW 3dB are doubled in volume.  The quiet parts in between the peaks are louder, but not so much that it sounds “compressed.”  Here’s an example of before and after using this setting.
Before:

After:
The differences here are barely visible, but the listener perceives the “after” audio as fuller, more balanced, with background instruments and voices more audible.

Using Audacity one can subtly shape the characteristics of any recording, and coming up with a pleasing balance is somewhat of an art.  Every engineer can even have their own ‘signature sound’ … which we can discuss in a later installment.

Robert Carlberg

6 thoughts

  1. I tried several other browsers and the images are not visible in any of them. I checked the source code and the images are linked to here: https://mail.google.com/mail/u/0?ui=2&ik=43d04a6b57&attid=0.1&permmsgid=msg-f:1847098287422593454&th=19a2361b270e59ae&view=fimg&fur=ip&permmsgid=msg-f:1847098287422593454&sz=s0-l75-ft&attbid=ANGjdJ8ZDhEqu9G9WUR1_VW7pUn2gT7fPBIAcx1rTvJEyI70lEO16jHulZcgxALN_kMhS_Yn8q4k33TXN3mjQ_9XVg69wu84oJ71PeuSSmye_8OvJDkHFwH3CtAaemE&disp=emb&zw

    Which looks to me that they are hidden behind a private gmail account. but I could be wrong I’m not a coding expert.

Leave a Reply to KyleCancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.