There are two mechanisms in CEP I use for this, AMPLIFY-'Hard Limiting' and using expanded view to see and highlight just the peaking sections to reduce just the peak section similar to what was already suggested.
If you have a lot of crowd noise sections, then 'Hard Limiting' is easier with simply amplifying everything by 5-12 dB to more normal levels, and letting the program automatically 'squeeze' all excessive peaks to stay below clipping. However, hard limiting does induce some artifacts that might be audible and sometimes adjusting the parameters of look ahead and timing of this feature can help minimize the effect.
If there are not to many places where the peak levels occur, then expanded view of each peak with -2 to-6 dB amplify reduction action of each small section has benefit of producing NO audible artifacts.
To be sure of this, I often use the 'zero crossing' feature to find where the signal crosses zero (sometimes placing a 'cue marks' there before and after the peak) and de-amplify that section. Finding the zero crossing is sometimes easier if just working on left, or right channels independently to find zeroes in each channel peak sections so less of the waveform is de-amplified so it is not audible for sure, then just apply regular normalization to whole waveform to get levels up for best listening experience.
I had to do the manual approach to recent Los Lobos concert by reducing peaks manual method by 2-3 dB and then normalizing. Used independent left/right channel zero crossing as most times peaks were ONLY on one or the other channel, but not always.
Manual process took about an hour of time. Results can be found at
www.sonicstudios.com/mp3_2slp.htm#los_lobos
Topic: How do I get a good level on a recording ? (Read 4275 times)