I want to express how much I love compressors,
but I make references to them as being satanic because too many people use compression without understanding exactly what is happening to the sound.
I will hopefully answer your questions.
Originally Posted by Raymond62
What do you think of using just a bit of compression to lower the highest peaks?
I would NEVER use a compressor for that purpose. The reason is that the rest of the music will be affected negatively.
If you set a threshold that is catching the excessive (over) peaks, it will also compress the music that is near the peaks. That is not accomplishing what you want.
Here's another way I look at using compression for this purpose, the user is too lazy to use fader automation to smoothly reduce the overall level, (mastering) so he sticks a compressor on the master, and is happy that the levels are no longer in the red.
A limiter on the other hand will allow the user to set a limit on the loudest peak, but here again, the drawback is that we can hear the limiter when the peak should be a peak and not squashed.
I recommend using fader automation instead of any plugin to take care of peaks. Too often, right after the peak, the compressor is either too slow to "come back up" or it is too fast and causes pumping in the low passages which is usually reverb. An MP3 example of this is in the Audio Mixing Tutorial I did for Gary Garritan.
SEE VIDEO AT THE BOTTOM OF THIS POST
Multiband compressors are fantastic. They are our friend.
Here is a hypothetical dilema, let's say we have some 1st violins in a piece we are mixing. When they play notes in their high register, they get on our nerves. Many professional vocalists often have a serious issue with this too.
What we need to realize (hear) in these times is that the vocalist or violins are not too loud, instead, a certain frequency is offending our ears. So, wouldn't it be nice if we had a compressor that was smart enough (like our ears) to reduce only the volume of the specific frequency range of the offending frequency. That's what a multi-band compressor is. Depending on how many offending frequencies are bothering us, we can control how much volume is reduced during these hurtful bothersome passages.
For instance, Digital Performer has a three band EQ. I typically set the crossover frequency between the three bands so that the middle band is only affecting the frequency range that I want compressed, that way the middle band will be responsible to do all the "ducking" of that shrill sounding vocal or violin/s.
In the following screenshot, you can see that I set the low frequency cutoff (crossover point) to 2330 hz., and the high end crossover point to 3450 hz. It is not critical to try to be precise in the choice of these frequency settings, just get it as close to the ballpark as you can. Don't ever ever get your panties in a wad being anal about numbers with all this stuff, just trust your ears. Did I say that?
You can see the violins' gain being reduced by the red level indicator "coming down" in the middle band. I set the threshold and ratio so that just the right amount of gain reduction is applied, in order to maintain smooth sounding violins.
I just updated this post with a video on May 02 2009.