We use an A&H Qu mixing board at church and the compression display there is 100.00000% logical to me.
The audacity compressor seems at best backwards or at worst totally nonsense.
I am an olde geezer so please tell me how to make sense of the compressor.
I need to run some tests to find better ways to reduce DR at church.
I was going to simulate parallel compression by compressing two channels differently then merging them in a variable ratio TBD.
But I have no way to tell what numbers I see may work in audacity really are so I could try them on the A&H mixer.
Threshold seems to be the level after compression not the threshold to start compressing at.
The ratio seems to work right but the display does not show the ratio starting at the threshold.
So please tell me what that display is really telling me.
And is there anyway to see it logically like AH does with their mixer?
I note in passing that Chris new compressor may be the bees knees but the numbers make even less sense and there would be no way to duplicate what I got from that compressor on the AH Qu32.
I alternate between gaffer, duffer, geezer, and fart.
Today’s Saturday?
Fart.
Chris new compressor
1.2.6 is the last known good “classic” one. There’s two tricks to making Chris work. Put a chunk of “fake” show before and after the “real” show. Chris doesn’t like sailing off the end of the show while it’s working. The look-ahead processor loses it. Cut off the extra later.
The other thing you can try is changing the first value (Compress Ratio, from memory) from the default 0.5 to 0.77. Don’t change anything else. If you do both of those things, Chris will simulate almost perfectly FM radio broadcast transmitter processing.
Chris isn’t a conventional “compressor.” It will adjust sound up or down.
You can freak it out. It doesn’t like dead air. It starts pumping background noise (like a transmitter).
I would rather just understand the picture with the classic one I have running.
It works well enough and has the right words just not the right diagram to match the words.
gaffer might work for an alternate. My Golf is good so duffer is out. OF occasionally fits:)
Chris compressor is good for what it does.
Unfortunately it does not match the problem we have at church that I need to use the Qu compressor to try to fix.
Time to try things there is limited so I wanted to simulate as much as possible on audacity first.
As you have observed, the graph does not tell the full story, and the Audacity compressor is somewhat different from a “typical” compressor.
The “graph” display shows input level on the X axis and the output level on the Y axis.
The Y axis is normalized to 0 dB output when the input is 0 dB - This matches the behaviour when “Make-up gain” is enabled.
Note that by default the threshold is the dB RMS level. If “Compress based on Peaks” is selected, then the threshold is the more usual “peak dB” level.
Thanks. That helps.
I wish they left off the make up gain as we do not always want that and do not expect it in the diagram until I tell it how much make up gain to use.
A&H shows two graphs - with and without the makeup gain to make it clear what we are seeing and doing.
I had the impression the folks doing audacity were going to switch to Chris Compressor or at best just keep the old one around not make changes to it.
My experience is the compressor seems good but the display to use it could be made more intuitive and it should give me the option for the amount of make up gain to use if any.
You can always turn off “Make up gain” and then apply the Amplify or Normalize effects with the amount of make up gain that you want.
I’m hoping that the updated version will be a lot more “standard” (like a “typical” compressor).
The updated version will have “real-time preview” (like the “Bass and Treble” effect) so that you can adjust the controls while listening to the preview.
I did not see any place to turn off make up gain. It just seems to default to -1 dBFS.
Note: I am on 2.0.6 which seems to be the last audacity for XP. Maybe later versions are different.
There is an option to add make up gain to 0dBFS which I left unchecked as I did not want any make up gain.
Unless this is another non intuitive reverse option where if I had checked it then it would NOT add the gain?
Just checked it and it added the gain when I had the box checked.
If it did not add make up gain I could always amplify to get any amount of gain added back.
What it should do is either nothing wrt gain, or else ask what level to raise the compressed result back up to.
It should never default to raising the gain to any arbitrary level on its own.
I also note that the compress ratio is confusing too. Instead of changing the ratio on the chart like A&H does it changes the threshold setting (or what I would expect would be the threshold setting) and the threshold slider moves the threshold as well as the start of signal.
Now maybe the chart makes sense which is why I asked the original question but to me and what I see on the Qu32 it is very confusing whatever it is audacity is really trying to show me there.
This is another weirdness of the current compressor. If you compress based on RMS and have “make up gain to 0dB” turned off, then the high level signal is compressed down with no make up gain. The weird part: If you compress based on peak level, then it becomes an “upward compressor” and increases the level of audio below the threshold.
Also at times it clips when it makes up gain.
Can’t say if it merely touched 0dBFS or how far it went past but the red lines show up as if it had made up too much gain at some spots.
Amplifying down seems to remove it visually but I do not know if it fixed it digitally or there really was no problem to start with.
I wish A&H had an upwards compressor but they are doing it in RT not batch like audacity does, so they only provide normal downwards compression.
I know that if make up gain is enabled, it does not clip.
As far as I’m aware, upward compression without make up gain does not clip either (I would be very surprised if it did, but I’ve not exhaustively checked).
You can check by zooming in close on the red lines. If there is just a single sample touching 0 dB, then it’s not clipped.
