Compression attack and release curve controls

[jithknot]

hey there
im making a compressor in matlab for my assignment in college
i just wanted to know if there are compressors with attack and release curve control for reference sake.

as of now the way my compressor works is that it compresses linearly. i.e ratio is proportional to the attack time and release time, which i think (correct me if im wrong) is how all or most compressors work.

all criticism and and inputs welcome :)

[Kurt]

Wouldn't a soft knee compressor work non-linearly, at least below the threshold?

[jithknot]

yeah it does work like that.

but the thing im after is what comes after it. and for now it works linearly, and it works for both attack and release., well i tried it, and i can see changes in the wave file, but cant really hear it, maybe because i don't have "golden ears" haha.

[chris p]

I'm not sure what you are asking, Jith - attack and release controls are generally time-based controls rather than amplitude response curves (which is what a ratio control and hard/soft knee controls are).

When you look at a comp's amplitude operating curve, you should see a 1:1 linearity up to the threshold, and then a lesser slope according to the ratio control from that point up (with suitable adjustment at the change over point if you have soft knee compression). However, this is a picture at an instantaneous point in time: given an input level, you can see what the corresponding output will be at that point in time.

Attack controls how quickly (usually in ms) the output will change from 1:1 linearity to that comp operating curve once the threshold is reached. You hear its effect in transients like pick or fingernail clicks on guitar, the snap of drum sticks on drums and cymbals, and the initial attack of all these sounds - they stay forward and present with longer attack times (because the compressor is still opening up as those parts of the waveform pass through it, and so they are relatively uncompressed even though the threshold has been exceeded), and more even and controlled with shorter attack times.

Release, similarly, is the time after the sound level dips back under the threshold for the compression to switch off. Its harder to describe its effect, 'cause to me it depends on whether your comp is adding its own "sound" (on top of amplitude control). Release controls have this sound fades after the threshold is passed on the way down (and hence emphasises or de-emphasises comp pumping effects). Remember its real effect is to keep the sound compressed even after the threshold has been crossed downwards for some length of time.

To get to your question, if you think of a graph showing compression (as an in/out ratio) over time as a loud signal is applied. The graph starts as a straight horizontal line at 1:1, and at some point drops to a horizontal line at 1:comp ratio after the threshold is exceeded, and then returns to 1:1 after the threshold has been whatever the opposite is of exceeded. Underceded maybe? Your question is whether the transition from 1:1 to 1:comp ratio, and the transition back, typically has curve controls? No, it doesn't. Typically its just a duration control, ie the duration it takes to get from 1:1 to 1:comp ratio. And when you think about it, it doesn't make a lot of sense to do anything apart from linearity here anyway. Suppose you had an exponential curve for your attack control. What would be the difference between a short or long attack time? Answer, not very much!!! They would both sound the same. Similarly for release time, you need to actually draw the time-based graph, pencil in some different curves and think what the effect would be as you fiddled the control - again, you'd find that you'd be mostly removing the effect of the control. So much so that you could in fact design a compressor where the attack and release controls are actually of fixed time duration, but the control varies the transition curve over that time, from log to linear to rev log. It would have the same effect on the sound as changing the time of the transition. I do think some comps are built this way, even if their controls are labelled with time graduations.

I hope this all makes some sort of sense, and as always I'm ready to stand corrected by those more educated in AE theory.

[heathen]

I think from memory the Sonalksis comp has attack and release curves, I could be wrong though as my memory is'nt that great.