Question

Some site seem to recommend normalizing recording to -3db but I am not sure it is a good idea since of all the side-effect normalizing can cause. Thanks.

Answer 1

I humbly preface my answer with the sage advice (not sure where I first read it) "edit with your ears."  In other words... the degree to which your normalize, compress, hard limit etc... depends a great deal on the "noise floor" of your track... and how it sounds when you're done processing... not what it looks like on the vu meter... provided of course that your tracks are not clipping.  

If your signal chain is relatively quiet and your noise floor down around -80 db then you can do the above without significantly degrading the quality of your demo, audition, finished product.

The goal of having your material end up at -3db is not a bad thing because that's what most people do... and if a potential client is listening to a dozen auditions and yours comes in at -12db... it might be a red flag... i.e. the client wondering why the volume of your audition is so low... and jumping to the conclusion that you're a rookie who doesn't understand gain staging or has a rinky dink home studio.

Another point... even though my noise floor is around -80db... I still do a routine "light" noise reduction process as the first step in preparing raw tracks so that when I add something like compression it doesn't noticeably raise the noise floor.

And a final important point in all this is the order in which you process tracks.  The following works pretty well for me.

1.  Normalize or amplify raw track to around -6db.
2.  use parametric eq to roll off the useless bass frequencies
3.  Light compression... 2:1 ratio that brings the level up to -3db
4.  back to parametric eq to sweeten to taste... i.e. I boost to bring up mids a little... and narrow cut for any problem frequencies that may make your track sound boomy or boxey.


Hope this helps... good luck.