A(nother) Case Against Linear Compression



Earlier in this Blog, I spoke about a case for using range-reduction linear compression ahead of Crescendo… Now in reconsideration, I am arguing against this practice.

Range reduction linear compression was applied uniformly across the entire audible frequency range, and was never intended for correcting hearing. Rather it was an attempt to bring up the faintest portions of the music, to make it a bit easier to reach for Crescendo, and minimize some of the sensitivity of the vTuning control for strong correction levels.

But that shrinks the entire dynamic range of the music. The loud portions in addition to the soft. While it sounded okay, with weak levels of compression, I don’t think that is really what is warranted here.

If we want to bring up the soft portions of the music, then we really ought to be just using NYC Compression. That’s what it was designed for, while leaving alone the forte sections. Furthermore, we really don’t have any trouble with the bass frequencies, just the treble range.

So… thinking about CLAS, or some variation thereof…

CLAS was originally designed, more than a decade ago, when it was common practice to record with 12-14 dB of headroom. Things were played much louder in those days.

Today, we have the international standard of EBU R128, which mandates that broadcasters should play back at a nominal 0 dBVU level = -23 LUFS. That is so much more sane…

But it also means that the old CLAS threshold level of -30 dBFS RMS is too close to the 0 dBVU level now. Whereas, -30 dBFS used to be about 15-18 dB below the average playback level, now that is only 6 dB below.

Secondly, CLAS was designed as a straight limiter (infinite compression ratio) and that might be a bit too severe for maximum listening pleasure.

So, how about a compression ratio of 3:1 in a NYC Compressor, with a threshold down around -50 dBFS? That is very easy to do… just change a few numerical constants in the code and recompile the DSP code. Voila!

Now we have a 2-band NYC Compressor with a threshold of -50 dBFS, a compression ratio of 3:1, and makeup gains running from minus infinity to +10 dB. (In a parallel compression arrangement, that becomes 0 to +12 dB boost at threshold and below)

That threshold is now 24 dB below the average recording level of the music, the compression ratio is more graceful, and we will generally use larger amounts of boost. Where we once used only 3 dB, we will more routinely use 10 dB. Tune to taste…

Sitting here and trying it right now. It really does sound good. But then, everything new always does…

This is a 2-band parallel compressor, treating bass and treble separately, with the dividing line at 725 Hz. Now when the bass is loud, but the treble is soft, I can bring that ppp treble up to the comfortable reach of Crescendo, and ff sections remain unaltered.

  • DM

[oops! He dropped his baton… maybe I should back off the pianisimo boost a bit? ]

Author: dbmcclain

Astrophysicist, spook, musician, Lisp aficionado, deaf guy

Leave a Reply