I am used to setting compression on an AH Qu mixer.
The terminology seems logical and works right for those devices.
When I try to compress in audacity either the terms are wrong or the compression is upside down.
What I expect normally is to see with OUT compression just a diagonal line at a 1:1 slope.
when I set a threshold and apply a ratio I expect that line to bend at the threshold.
What I am seeing on 2.0.6 (for my old XP) is that as the threshold goes up so does the signal and no compression happens until much later. This , to me and how I think , is bizarre.
I see no changes when I adjust noise floor and that makes me wonder just where the floor really is.
Do I have to normalize the signal and then compress accounting for that gain when I set parameters?
I can play with things and sort of get a compression that is ‘okay’ but I would rather be able to dial in directly all the values and know it will do what I asked for it without guessing.
I also have the compress dynamics (is that Chris’ compressor?) which is even more bizarre as I have no idea what the result would be based on the parameters I can set. So even more playing and guessing. Might be okay for batch but if one needs to compress in RT then it is not so useful.
So, any advice on helping me wrap my head around how audacity views compression vs how the real world sees it would be appreciated.
The default attack settings on Audacity’s native compressor is ~200ms.
For some applications this is waaay to slow.
There are free VST plugin compressors which work in Audacity which work in real-time,
so you can play around and quickly get the effect you’re looking for, e.g. …
Thanks for that tip.
Attack time is another issue as is release time which is not useful in the ranges allowed.
I may need to go to VST but have been avoiding that, as using the NY plug ins has been easier so far.
I think I might find it easier to write a correct compressor that is better!
But I would like to defer that task for a while if possible.
What I need is to understand how the alleged settings actually change things as they are totally different than the ‘professional’ QH Qu32 mixer I am using at our church. On it all the things like threshold make sense and stay put where you want them. The audacity compressor is illogical and makes no sense the way I know the words being used.
With the audacity compressor the threshold moves when I adjust the ratio which is counterproductive.
The ratio should bend the output/input curve where the ratio starts being applied.
When I adjust the threshold it moves but not logically.
The threshold seems to be 60 minus the real threshold. But then the ratio does not make sense as it is shown.
When I set the threshold to -14 the diagram says it is really at -42 for signal start and compression actually starts at about -4 but is hard to tell on that picture it shows.
That’s the one. I don’t think it’s useful to try and change Chris’s settings. Just to know the metaphors and results.
He designed the compressor so he could listen to opera in the car. If you have an even moderately noisy car, you know that’s not fun. So his look-ahead compressor “knows” when the scene is going to switch from tutti orchestra to one breathy soprano way out in the south forty and adjusts volume accordingly. Very well. Chris was doing this upside down. He knew exactly the outcome and adjusted the code accordingly. Not design a generic compressor and work down to the application.
But I fell in love with it because of it’s ability to simulate Broadcast Compression. I used to enjoy taping “Car Talk” from the local FM station, burn it to a CD and listen in my truck. Scene shifts to them putting the shows on-line. Crisp quality and no more FM hiss. Yay!!
However, no more transmitter audio processing, either. It turns out one performer mumbles in his beer and the other has a thermonuclear laugh.
Chris to the rescue. I move the first variable (compress ratio??) from the default 0.5 to 0.77 and don’t change anything else. The result is an exact match to the KPCC radio transmitters — No annoying volume swings and no hiss.
There is one way to freak Chris out. The latest version (1.2.6, I think) didn’t like running off the end of a show. The look-ahead could be fooled. That never bothered me because I always sliced off both ends of the show anyway, but if you need every note of every song, put “something,” some sound at both ends so Chris will have something to chew on and then slice it off later.
This is not likely to be fixed, because Chris reached end-of-life.
I had picked some random ones and the results were an improvement for sure!
Could you possibly help explain the way the older compressor that shipped with audacity works?
The settings make no sense they way the diagram moves. And they totally differ from how A&H does it on the Qu32.
And the audacity usage is different from how I know the words, and as some textbooks use them.
I will check out the ones you suggested.
I really should get up to date on using VST:)
I know the advantages just been procrastinating about learning yet another new thing.
I had not planned to use RT but had wondered if it could work like for podcasts.
Still would be nice if the compressor worked like the rest of the world so you could use it directly in batch
and not have to play with it to see what it will really do.
Could you possibly help explain the way the older compressor that shipped with audacity works?
No. I have no idea. I can probably guess what they’re doing by the graphics and instructions, but I have no actual touchy-feely experience with them solving real-world problems.
That’s another advantage of Chris. Input volume is open-ended. My only experience with the Audacity compressors wasn’t the best because as near as I can remember, you have to pre-condition the input. You have to get close first, before they will do anything useful.
I designed the original Audiobook Mastering Suite and it, too, is open ended. You can throw floor sweepings in there and the performance will come out passing the first two technical standards, Peak and Loudness. You can’t submit floor sweepings for publication (it will never pass everything) but you don’t have to pre-condition the input volume, either. That’s one decision a new user doesn’t have.