Normalising isn’t the right tool for matching the perceived loudness of a bunch of audio tracks. All normalising will do is find the level of the highest (peak) amplitude within your audio track and then amplify the whole audio track by the difference between the peak amplitude and the desired reference. As an example, if the peak amplitude is -5dB, and you normalise to -0.1dB, the overall gain applied to the whole track will be +4.9dB.
The problem with this approach is it takes no notice of the actual content of the audio, if you like how much energy is in the music. You could have an audio track that has a peak amplitude of -10dB, but it will sound louder than another track that may have a peak amplitude of -3dB. Were you to normalise both those tracks to -0.1dB, the audio track that was originally at -10dB will be excessively louder than the other audio.
The cheapest tool you can use for matching audio loudness levels is your ears. Play the tracks back and increase the gain on the tracks you feel are quieter, and decrease the gain on the tracks you feel are louder. Chances are, if you feel the audio seems relatively matched loudness wise, so will your intended audience.
There are software tools that perform very good analysis and loudness matching, but come at a cost. I use Adobe Audition to match loudness and import into PSP8 as it does a better job than PSP8’s built in loudness matching. That’s not to say the PSP8 loudness matching is no good, far from it! I just find the Adobe Audition tool much more flexible in its use. For most folk, if you’re on PSP8 or are thinking of upgrading, the built in loudness matching function will be fine. I’m fortunate to have access to better tools so I use them.
For those curious about the debate regarding whole loudness matching, the following link
http://www.soundonsound.com/techniques/end-loudness-war is a good place to start.
Steve