k y m a • t w e a k y — the kyma collective || view the current website || February 2015 Archive

/ WebHome / Sounds / Share.HandsFreeCompressor


Share Section



All Sections


Login / Register 
Change password
Forgot password?

Symbolic Sound

Kyma Forum
Eighth Nerve

TWiki Links

TWiki Formatting FAQ
What is TWiki?
%SESSION_ENDIF% twiki.org
Attached Sound is a new design for sidechain management of a compressor. At issue here is one of wanting detail in the sound by having a fast attack, and a psychoacoustic release, but not wanting excessive pumping in the output of the compressor...

I discovered last night that the average absolute value of a signal tracks about 2 dB below the RMS level. But averagers take a long time to respond.

At the same time, a fast attack peak follower with a relatively long release runs about 6-10 dB higher than the RMS level.

Using either of these two alone produces problems related to either failure to compress sharp transients (in the case of slow attacks with the averager), or too much pumping, requiring long release times (in the case of a peak follower).

But when combined together, in parellel, they track quite well at pretty close to 6 dB above the RMS level. So you simply divide their sum by 2 to get the RMS level. The result has the smoothness of an RMS compressor combined with the fast attack capabilities of a peak compressor. You can play with the release times, and there is some variation in the resulting sound, but not as great as with either one alone.

This is a hands-free compressor for general use. The Peak Detector uses a 0.01 ms attack (instantaneous!), and a 30 ms release. The averager uses a 30 ms time constant. [The averager is just a peak detector with equal values for the attack and release parameters.]

Using this design, I can't believe the detail that I get to hear in the music, while at the same time, assuring that no sharp transients go unpunished...

[it is psychoacoustically related by virtue of the choice of 30 ms for the release times. Our nervous system takes about 30 ms to recharge fired neurons. This corresponds to the duration of temporal masking, where sounds following a sharp transient go unnoticed unless they occur more than this period after the initial transient. With this design I get good compression and I still get to hear the details in the reverb tails...]

-- DavidMcClain - 24 Feb 2004

© 2003-2014 by the contributing authors. / You are TWikiGuest