Part of what a brickwall limiter does is to effectively chop off the top of a waveform's curve, then match the signal level to a preset ceiling that is generally set to be below 0dBfs (ie maximum possible sample value). When this clipped waveforms is eventually played it gets converted to an analog signal. Depending on the specific analog signal chain (and DAC) used, power supply etc, you tend to get a degree of apparent analog reconstruction - ie where the lost curve gets recreated due to resonance/limited slew rates in the analog signal chain - ie the analog level overshoots the the level that should strictly result from the digital sample values. With poor power supply headroom, you get other very audiable problems as well.The Carpet Cleaner wrote:Why can't you normalize? You already normalize when you limit your track and hit the ceiling. What happen if yu normalize it after that?Pepehouse wrote:Limiters/maximizers are very transparent today music doesn't sounds bad if you use them carefully. Download this and put it in your master channel, it's free: http://www.yohng.com/w1limit.html
I use this settings for my DJ mixes but them should also work on a single track: Ceiling=-0.2dBs Threshold=-6.0dBs Release=40ms.
If you like what it does to your music think about buying a good one like Voxengo Elephant or Waves L3.
And never normalize.
So, when applying a brickwall limiter, you typically set the limiter to clip at a little below the 0dbFS threshold -0.5 to -0.3dB being a common range of settings so that the overshoots generally stay under 0dbFS.
If you were to normalize the result back upto 0dbFS, then when the waveform reconstruction occures in the analog domain, on poorer equipment with little or no head room (mp3 players, ipods etc sufffering worst from this), then the reconstructed waveform can clip very audiably, especially if battery charge is a bit low. Also as mp3 encoding in some ways represents a digital version of analog reconstruction depending upon the specific pre-processing employed by the mp3 encoder, then normalizing clipped waveforms back upto 0dbFS can make a right mess of the resulting mp3 resulting in very audiable distortion on pretty much any player.