Compressor insides

DSP, Plugin and Host development discussion.
RELATED
PRODUCTS

Post

For what its worth (likely very little) the behavior of that modern comp looks about like most compressors I ever tested back in the old days, with test signal played back from tape recorder and signal response plotted with a Tek memory oscilloscope running a real slow sweep (so as to see several seconds of audio envelope frozen on the tiny glowing green scope screen).

Then I could place a small machinists ruler up against the screen, measure the amplitudes and times, and attempt to guestimate attack and release settings, ratio, etc.

I think one might have to go to a lot of trouble to make a conventional analog compressor that does not behave similarly to the modern comp. Am not saying analog is better, or that some other digital behavior might not be an improvement. Just that a conventional simple analog compressor with attack and release will tend to behave like that modern comp. Am pretty certain any digital compressor I ever made would look like that as well. Haven't made one in awhile, and the last incarnation was direct-x plugin format.

Usually that "release tag" where gain gradually creeps back to unity after the signal falls below threshold-- In my testing the test tone had below-threshold signal both before and after the loud bursts. Otherwise, I wouldn't have been able to view the release tail with the oscilloscope, if looking from the outside at a black box. If the test tone's last loud burst would have ended in silence rather than a final low level tone segment, I would have needed to find a suitable location inside the box to probe the envelope shape. If I wished to view the below threshold release behavior.

However, probing the innards of an analog compressor made by someone else might not be productive. One never knows about nonlinearity of the components, and sometimes the final envelope inside could be current rather than voltage, sometimes making it difficult to find a good circuit point to probe.

Post

here is a picture of the compressor with the fastest attack

also i just noticed there is a sidechain parameter that acts like a HighPass (and you can choose the frequency) so i suppose that's the clue here.

Image

@PurpleSunray
i think i get it now : the reduction amount is a RMS, so when there is a peak in the input, it takes some time to full get to the "ratio" compression amount. is it it?

but as you can see on the modern Comp, the curve has changed with the attack, so here, it's sidechain-based i guess.

i'll post other curves of other compressors once we're done with this one. Could be interesting: it would show which compressor is more unique than others.

Jeff

Post

here is Reaper from Cockos:

Image

Post Reply

Return to “DSP and Plugin Development”