Gonna try to do this without uploading pics. Let me know if they’re needed please.
I’m experimenting with compressing a recording of voice and guitar with a high amplitude variability. There is one spot where the voice gets too loud and pierces the ear at my overall target volume.
I am using the built-in compression effect, not an add-on, and I am not using the make-up gain setting, as I plan on setting the overall level later.
I used the peak amplitude analyzer on all of the audio right around the ear-piercing peak (but not the peak itself) and found that peak amplitude was -7 dB. I analyzed the part with the ear-piercing peak, and found a peak amplitude of -3.5 dB. So, I selected the whole section, peak and quiet part, set my compression threshold for -7 and started playing with the ratio.
No matter what I do, the volume of the peak is not changing. I did notice, however, that a portion of the non-ear-piercing part that is well below the threshold (I measured it’s peak at -14 dB) is reduced in volume when I apply the compression. Toggling back and forth between compressed and not (using ctrl-z/y) reveals no visual change in the peak, but a reduction in amplitude in the quiet part.
Isn’t that backwards? I thought the compression was supposed to affect only the amplitudes above the threshold, and have no effect below. What’s happening for me is the opposite!
You may need a faster attack time. If it seems to be working “backwards” (expanding instead of compressing) it could be a timing issue.
A higher compression ratio and/or higher compression ratio should make a more dramatic effect too. You can also try checking “Compress based on peaks”. And, you might want to normalize for 0dB peaks before using compression.
I’ve tried it with the fastest attack time the tool allows - .1 sec - with no change. The only thing that’s worked (since my original post) is to select only the peak I’m trying to compress, and nothing around it. When I do that, it works as expected. As soon as I apply the compression to a selection that includes softer areas directly before the peak it stops functioning as expected and starts lowering the soft part instead of the loud part.
Bit late to the party, but it points to an issue I’ve seen. When compressing an entire audio track, I see the fade-in you describe, even when I start with 5 sec of room tone, and at the other end, I see trailing room tone get amplified.
My “solution” is to add fake text, then chop it out afterwards, but is there a more elegant approach?
