I have tried to reproduce it in a different module but it didn’t work for some reason, both compressors were working about the same way without following my predictions. I had the signal peaking at 0dB on master using a gainer (since there is a 6dB headroom on tracks) and tried to bring it to -2dB using either of the compressors with a -4dB threshold, 2:1 ratio and a very short attack.
Instead this time they both put the peak to about - .3dB, so now I am just posting a pattern from the module I was working on where the issue consistently manifests itself. I had to convert the samples to ogg format to save some space, it changed the peaks somewhat but the issue is still obvious.
The above tool is freeware, very useful btw, and @ sonalkis.com ( spelling error maybe )
Using the tool above, I was able to decipher that you probably need a look ahead limiter. I suggest the freeware version of Event Horizon… ( if it still exists. ) in the Reaplugs pack, at Reaper.fm
I’m using the payware, Event Horizon… Good stuff. I’ve had it for years, and you just surprised me again with, “how good it is, and how worth the money it is.” I don’t recall the freeware one with VU meters, but I could be wrong… Download the demo, or whatever, and you can hear what it does to your track…
btw, the sonalksis thing is freeware… it shows rms, its really precise. i’m using it all the time… cause a lot of times my volume goes below where renoise shows me. cause i mix very quiet… there is no need to, “hate on the freeware.”
Why not push zero? As long as it’s not clipping or altered too much by the compressor why not? Cutting peaks no one can hear is fine by me, I don’t see the problem. The false dichotomy loudness war guys create is ridiculous. It doesn’t have to be squashed or no compressor at all.
What? Who’s hating freeware here? I don’t see what you’re getting at.
As there is an Attack time I wouldn’t expect your Peak readings to be affected that much. Transient hits with a peak, after x amount of time the level is lowered, but the initial peak level has still been allowed to pass.
I think I have finally figured it out. It appears that the ratio setting behaves strangely.
I put a 0dB sine through the compressor with the initial settings of -20dB, 2:1 with whatever attack/release. Now, according to the ratio (or to the way it is commonly understood), 20 dB of the signal over the threshold should produce a -10 dB peak. However, it is not the case (look at the continual peak, the other values are contaminated and irrelevant):
3444
As we can see the result is only a 2.54 dB reduction of the signal. In order to get the sought after -10 dB increase, one should the ratio to about 11:1 Don’t ask me why, but that is how it works. Now, coming back to my original post, if I want to have the threshold at -4 dB and have the reduction of the peak to -2 dB, it appears that I need to set the ratio to 9:1
3445
If anyone would be willing to figure out the formula one needs to use to interpolate to conventional ratio values, that would be awesome.
However, when I came back to the original mix and tried to apply this newly acquired knowledge it still did not work at all!
Turned out, that the compressor does not like stereo signal with a lot of difference between the left and right channels. Apparently, it seems that if one channel goes over the threshold while the other stays below, very little compression (or none at all) is applied to the signal so one ends up shooting over the top again.
But if one converts the signal to mono first, the compressor works just fine in a predictable fashion described in my previous explanation.
Everybody, thanks for your help and participation.
There is still things left to figure out about the maximizer, but I think it might be left for some other time:)
Reminding myself to give this a shot. I think this is really cool. Btw… I think I have missed the whole point of the thread earlier… It doesn’t matter really, but I’m glad I saw this when re-reading the thread.