Quote:
i see what you done here, but there is problem. You'll have bigger "overall difference" in RMS level between highly compressed tracks and dynamic tracks.
Could you tell me a few tracks such as that to test with?
Quote:
You done it with slower "UpSpeed" but that's not the way. usually AGC works in ratio inf:1 which will give best AGC riding. ST is inf:1 in AGC. So, for this what you want to do, you need adjustable AGC ratio.
The gating brake I did works for tracks that start out soft and should stay soft for a while, as well as tracks like The Pretender where there was a recent very loud passage followed by a brief quiet passage before resuming being loud. It does not work for the "torture test" file though, as "Dreams" is a quiet track in comparison, and there's a fade-out involved too. The spoken intro to "When I Grow Up" should be at a lower volume, but it isn't.
The problem I see, which is what I mentioned in the wish list thread, is that the gain is never reduced on a track change or a silent passage. The only time gain gets reduced is if the look-ahead determines that what's coming up is too loud. This means that there will be instances where the soft passages have too much gain applied to them, and are thus "too loud" themselves. If the soft passages are too loud, then the transition to a loud passage will be more pronounced, as the starting volume will be higher than what it should have been.
I did some brief research (like 30-60 minutes worth) into AGCs, and one thing that was mentioned is that they can tend to make the softer passages too loud and the loud passages too soft. I also saw something about having two control variables, one on a short time horizon and the other a longer time horizon.
I really think the gain needs to be reduced more than just for times when the volume will be too loud. I understand that it "could" lead to pumping, but not if you weight the gain reduction, making it a conservative reduction.