"Normalization" is a mathematical concept related the peak values in the file. As you know, normalization doesn't correlate well with perceived loudness. There are lots of normalized quiet-sounding recordings.
What you want is "Volume Matching".
If you are making an album or otherwise working with a limited number of files,
it's best to do it by ear.
If you want to do it linearly (without additional compression) the best approach is to normalize ("maximize") all of the tracks first. Then choose the quietest-sounding one as your reference and adjust the others
down to match (as needed).
Or with compression/limiting (with make-up gain) you can bring-up the quieter songs.
--------------------------------------------------------------------------------
If you want to volume match your entire music library -
ReplayGain works with your player software at playback-time to adjust the volume. The file is "tagged" with a ReplayGain setting, but the actual audio data isn't touched. Apple's Sound Check is similar.
<a href="http://mp3gain.sourceforge.net/" rel="nofollow" target="_blank">
MP3Gain and
WaveGain use the same volume-matching algorithm, but these "permanently" change the volume in the file so they are compatible with any player (or you can burn a CD, etc.).
ReplayGain (and related) have an "album" option that adjusts the album as whole, leaving the quiet songs (relatively) quiet and the loud songs (relatively) loud so when you play the whole album you hear it as intended.
Since many quiet-sounding songs are normalized (maximized) you can't simply boost the quiet songs...
You have to make the loud songs quieter. ReplayGain (and related) use a default target-volume
that makes many (most?) songs quieter. Some people are disappointed by that, but of course if you have enough analog gain you can listen as loud as you want.