From what I understand ‘normalize’ is for an individual track’s sound levels and can’t be used to match multiple songs together.
Right! Normalization works on the peak, which correlates poorly with perceived loudness. Many quiet-sounding songs are normalized/maximized for 0dB peaks.
And if you normalize an album/EP as a whole, the whole album/EP will be adjusted by the same amount (not to the same amount) and the level will be based on the highest peak in the whole album/EP.
If you have a “manageable” number of tracks, it’s best to do it by ear -
1. Normalize all of the tracks (individually) for 0dB peaks. That makes all of the tracks as loud as possible without clipping (without distortion). (Some people prefer normalizing -1dB or so.)
2. Listen the the tracks, and if they don’t sound equally loud, choose the quietest-sounding track as your reference.
3. Use the Amplify effect to adjust the louder tracks down (by ear) as necessary for equal perceived loudness.
\
If you want an automated approach or if you want to match the loudness of your entire music library, look into ReplayGain, or the related MP3Gain or WAVgain, or Apple’s Sound Check, etc. There is also a [u]ReplayGain Plug-in[/u] for Audacity. These tools try to analyze the perceived loudness, rather than simply looking at the peaks.
But since many quiet-sounding songs are normalized/maximized and can’t be made louder, you’ll find that these tools make your music library quieter overall.