More commonly a brickwall limited does have a finite (non-zero) attack time to reduce the distortion from amplitude modulation. The way you get exact brickwall limiting with a non-zero attack is that you use a look-ahead window and then use a finite-time attack towards the maximum value currently inside your look-ahead window (which is why you need sliding max). If your attack time then is equal or less than your look-ahead window length, you will hit all the peaks in time.Miles1981 wrote:brickwall = attack = 0ms, usually peak (because you want the peak to be limited, not the RMS)
analog = attack > 0ms
Just doing the above won't necessarily result in a very competitive limiter though, so in practice you probably want to work on multiple time-scales at the same time, probably doing some form of more traditional high-ratio compression for the overall gain and then treat the peaks separately using a small look-ahead window.. but like.. that's the basic idea.