I used Sound Forge on my first demos a few years ago. It took me awhile to learn, some trial and error whether to choose the peak value or the RMS and then scanning the song and diddling around a lot with things like dithering I think it was called.
http://ScottLMusic.com/Listen

A few months ago I was putting up some demos of me and Chuck. I tried Sound Forge again and it was hit and miss again. I guess with some effort I would have gotten it under control again, but I didn't want to put in a bunch of time doing it. Hoping I'd find a quick fix I did. I used Audacity a free recording software with all the plug-ins a guy should need. It has only a couple choices for normalizing, so I checked the appropriat box and normalized all the tunes. To me they all came out pretty good except a couple which I manually set the volume control in Sound Forge, I think it was.
http://ScottLMusic.com/Chuck_Wheeler

The problem with this normalizing is it depends on all the factors in each recording. If they are done with a similar setting and sounds etc it is easy to do and fine. But when you've got something different going on like when I did Take The A Train as a big band number with no vocals, the peaks are different.

MP3 Gain is another free program that will normalize mp3's only and it takes into account that some songs are perceived as loud or softer when in fact the waveforms tell you something else. I know DJ's like to use this on their thousands of mp3's.

I'd sure try the free approach first myself

Scott