And (when mastering) the results of parallel compression ARE often the same as the results of upward compression. The difference is that parallel compression is more refined with respect to control of the effect. With typical low mastering ratios and low threshold, parallel compression is the same as upwards compression. That said, parallel compression offers more variation, as typical upwards compression is not always suitable.
As Bob Katz says in his review of a Weiss device: Concerning upward compression, I feel that the ear forgives the raising of soft passages more easily than lowering of loud passages. This new software feature, also known as parallel compression, is engaged in the \"gain\" menu, putting the compressor section in parallel with the \"dry\" signal, and achieving an entirely different sound. If delicately performed, parallel compression can be extremely transparent, especially useful for classical or acoustic music. Distortion is reduced by several dB, since part of the signal passes through unprocessed. In essence, adjust the amount of compression by the gain makeup control instead of fiddling with thresholds or ratios. I recommend presetting the compressor's threshold to -50 dBFS, ratio to 2:1, attack to 20 µs, and release to 300 ms or so and leave them alone - simply adjust gain makeup to taste. During musical passages, the compressor section is in gain reduction nearly all the time. As the sound gets louder, gain reduction increases; thus the compressor contributes very little or no sound during loud passages.
Conversely, as the sound gets softer, the compressor raises gain because there is less gain reduction. By using an extremely fast attack (and the built-in look-ahead), transients are preserved, unlike the case with standard (downward) compression. The parallel compressor can be set to a frequency band, turning it into a dynamic equalizer. For example, by confining it to the bass range you can fatten a bass instrument, or to the treble range, which gives presence without making harsh sound at high levels.
Upward expansion, also known as the \"uncompressor,\" was added during a recent software revision, and in version 2.0, the range of ratios was made more useable. Simply set the ratio to less than 1, changing the unit to an expander. Remember to think \"backwards\" when using this feature; for example, fast attack times increase apparent transients. It is not possible to restore dynamics to hypercompressed material, but you can add more life to material that has been just a bit overcompressed. However, if overdone, upward expansion can sound as bad as poorly-done compression. For example, low-level passages start to get lost, or sounds seem to \"bounce.\"
Dynamics processing becomes a serious art when using upward expansion in conjunction with parallel compression in another processor (you can always get another Weiss!). Two threshold points hinge the material, one to give a little more life to crescendos and another to prevent the softest passages from being lost.
and, further comments in 1999 from Katz, Gerzon and Seva:
From: Bob Katz (
bobkatz@digido.com)
Date: Monday, November 15, 1999 10:45 AM
This technique is designed to compress with minimal or no effect on transient peaks. Mix the source with a \"parallel compressor\" (which some people call a side compressor, but this term is confusing). The source must be delayed to match the delay of the parallel compressor and avoid comb-filter effects. To determine the delay, invert the polarity of the source or the output of the compressor, set the compressor's threshold to 0 dBFS and makeup gain to unity. Mix the two and adjust the delay for complete cancellation. This delay will usually be a few samples.
