Three waveforms, top to bottom: Audacity compressed, Reaper compressed, original.
Notice the threshold and ratio in Reaper are waaaay more severe than in Audacity, yet the Reaper result is much lighter (meaning, much closer to the original) than in Audacity. I want what Audacity is giving me. The peak between fifteen and sixteen seconds in the Reaper-result is especially bothering me.
I'm not even sure the questions to ask. What am I missing?
400ms is an absurdly long attack time, and 1 second for release is pretty crazy too. ReaComp is giving more makeup gain because the threshold is lower and the ratio is higher. Either of those alone would cause more automatic makeup, and both together just compounds the issue.
400 ms atk time seems excessive. I can't imagine many cases where I'd want to use that long of an attack time, unless I was doing some kind of RMS averaging stuff.
>>EDIT<<
I wonder if the audacity "sec" should actually read "milliseconds". Seems off to me to be working in seconds for atk/release parameters. Try using 0.4 and 1 millisecond atk/release times and see if you get closer to what you hear in audacity.
Try using 0.4 and 1 millisecond atk/release times...
That would be extremely fast and would probably cause distortion on most sources. It does seem very odd that 1 second is the minimum release time!
According to this https://manual.audacityteam.org/man/compressor.html it looks like the attack is some sort of “pre-attack”, but then the release clearly starts after the input drops, so it’s not the same as ReaComp’s pre-comp. Overall that Audacity thing is just really strange.
As others have said:
..3:1 vs 7:1
..different methods of attack/lookahead behind the hood are entirely possible (audacity isn't constrained to realtime)
..Audacity doesn't list the RMS, which you have set to 0 in Reaper. 0 means 1:1 peak response, increasing it will round out the response to have more effect on longer areas of loudness vs quick spikes. So we have no idea what Audacity is doing there.
Uncheck the make up gain button and they'll be closer, of not very close. That's normalizing it up to 0 db which the Reaper example isn't doing. You could easily do it with Reaper, just not within Recomp, by double clicking on it (or going to the "item properties" menu and clicking normalize. If you do I would back it off a few db as any further processing might clip it and there's really no point in normalizing early in the game anyway.
Is this a spoken word recording? As mentioned, the compressor settings of .4 and 1 sec won't give best results. .4 is too long attack and 1 sec is much too long release if it is. Start with standard voice over compression of 20 to 50 mill attack and 150 to 350 mill release, keeping the ratio there at 3:1. If it reduces peaks by 3 - 6 db and you feel it needs more, best to add a second comp/limiter after instead of adjusting that one to hit it harder. Set the second compressor to a high ratio 10 or 20:1 make attack and release slightly faster and set the threshold so that it just hits the most transient places and takes them down 2 to 4 db. You can go more compressed on both if it doesn't start sounding too squashed.
You can check the auto make up button in the Reacomp(s) to assure the level ends up at least where you started after being compressed. Or you can do what Audacity is doing and normalize it.
__________________
The reason rain dances work is because they don't stop dancing until it rains.
That Audacity compressor is doing things that aren't possible with a real-time comp like ReaComp.
Overall, it's pretty strange and I'm not sure I like it.
0.1 s attack, 1 s release, 10:1 ratio, -6 dB threshold
Source, Audacity Comp, ReaComp
That Audacity compressor is doing things that aren't possible with a real-time comp like ReaComp.
Overall, it's pretty strange and I'm not sure I like it.
0.1 s attack, 1 s release, 10:1 ratio, -6 dB threshold
Source, Audacity Comp, ReaComp
Hmmm. I don't like the look of the audacity plot. Must be doing something really strange, it doesn't look like the response of any compressor I've come across.
Wow. So many replies! Thanks to everyone for all this.
I've been completely swamped and have not been able to deal with this. It is important (critical) that I figure this out (and obviously I've betrayed my ignorance about the ins and outs of compression ). I will take all this advice into account as soon as I have a moment to do so.
I should say, however. I started Reacomp with the same exact settings as in Audacity (as much as is possible). Only when I got such a different result did I start to ramp it up to the more extreme settings as in the above screenshot.
Based on ErBird pictures, it seems that Audacity used some look-ahead. In ReaComp, you can set the precomp value to mirror that.
Maybe that helps in addition to all the other advice you have already received.
Yes, that must be it. However, the attack/release curves seem to be completely linear, which is weird. To the OP; try a 100ms pre-comp time, a 100ms attack time and a 0.9 second release time, and see if you get any closer.
Based on ErBird pictures, it seems that Audacity used some look-ahead. In ReaComp, you can set the precomp...
It might help some but ReaComp’s lookahead also affects its release because it’s actually just a negative delay in the detector path. Audacity’s release is not like that.
It might help some but ReaComp’s lookahead also affects its release because it’s actually just a negative delay in the detector path. Audacity’s release is not like that.