PARALLEL PROCESSING

ALEXANDER MAGAZINE

Motown Heritage Mastering Seminar #4

June 18, 2011 - R. Dennis, Instructor

Parallel Compression

The illustration below shows the desired result of pop mastering compared to what is obtained with a mastering "maximizer" alone. In mastering for pop, we want to increase the loudness by making the sound denser. We, however, want to make all recorded sounds evenly dense. A maximizer works by squashing the loudest instruments, killing any sense of dynamics for these instruments.

Parallel compression gives us a way to use compression to more evenly "densefy" the mix. It's pretty much the only technique which directly brings up low level sounds. Parallel processing is the mixing together of two versions of the mix that have been compressed differently. It can be mixing together an uncompressed version of a mix with a compressed version or can be mixing together a lightly compressed version of the mix with a more heavily compressed version. Each "version" of the mix would be called a "stem."

A key to understanding how parallel compression can accomplish this, and a key to setting up parallel compression, is to understand how signals of different levels add together. The chart below shows this.

                SIGNAL ADDITION CHART
[Key: Signal Level Difference | Maximum Addition |   Minimum Addition]
         0 dB                                          6 dB                            3 dB
            1 dB                                         5.5 dB                         2.7 dB
            2 dB                                         5.1 dB                         2.5 dB
            3 dB                                         4.6 dB                         2.3 dB
            4 dB                                         4.2 dB                         2.1 dB
            5 dB                                         3.8 dB                         1.8 dB
            6 dB                                         3.5 dB                         1.7 dB
            7 dB                                         3.2 dB                         1.6 dB
            8 dB                                         3.0 dB                         1.5 dB

Let’s say that we are mixing together a compressed stem and an uncompressed stem. Let’s say that we have the threshold set to make the compressed stem make 8 dB of gain reduction on a loud sound. So the compressor reduces gain 8 dB on this lead instrument peak but doesn't decrease gain on the subtle sound that follows it (because the level is below the threshold level).

In this example the subtle sound on each stem are at the same level (no gain reduction), so when you mix the stems the level of the subtle sound increases 6 dB. For the peak of the lead instrument, however there is a difference of 8 dB. The compressor brought the peak on the compressed stem down 8 dB, but the uncompressed stem has that peak at full level. The loud lead instrument peaks of the two stems only add up 3 dB. Since the lead peak added 3 dB and the subtle sound added 6 dB, we have effectively raised the level of the subtle sound.