I made a little audio essay to show, more-or-less, how my SpinTunes 2 shadow song "Sherman's Lament" evolved. Note that this whole track is not compressed or limited until the end, in order to show what compression and limiting is all about, so it has a very wide dynamic range; you might want to listen on headphones and be prepared to turn it up or down for some very loud and quiet bits.
It would be nice to imagine that I wrote the complete vocal, wrote the complete melody, recorded a reference track, and started recording parts for it, each of which was perfectly usable as-is and which I never had to revise or change, and Joe and Denise's contributions fit perfectly and also didn't need any revision. I'd also be happy if I could say that my own vocal and instrumental parts worked out beautifully with a minimum of takes and my performances were stellar.
Of course, it was not quite this straightforward, and there was some thrashing and revision. I'm not the best guitarist or singer under the best of circumstances, but past about 2 a.m., the need for sleep starts to take its toll and both of these things get pretty grim; I can't keep my guitar parts tight against the click track, I forget what chords I'm playing, I can't remember my own lyrics, and my vocal pitch wanders. Mercifully, I can't remember every detail of everything I had to redo and revise, but you can read about the process here in one of my other blogs, Geek Versus Guitar.
The clips are roughly in chronological order. First, we have some early lyrics for chorus and what was going to be the bridge, with very rough drafts of the melody, that I sang off the top of my head into my digital recorder. I wound up not listening to these again, but the next step was to compose a melody in MIDI note form and sing with it. There are some examples of these scratch tracks I recorded for reference -- either for me to sing against, or for Joe or Denise to sing or play against. I'm including only a small subset of the actual bounce files I made during this process and none of the test mixes -- there are almost 20 of these drafts!
On this song I had help from Joe "Covenant" Lamb (male voice) and Denise Hudson (Rhodes piano, synthesized violin, female voice). These are in their raw form with no processing, and at their original volume levels. Note that Joe sends his vocals already compressed with reverb because that's the way he likes it, and Denise sends it unprocessed, which means that it needs compression and EQ before it can go into the mix, but which lets me use my own discretion as to what to do to the tracks.
After those clips, there's a section that shows how these parts are situated in the mix: what their volume levels are, and where they are panned. The parts are repeated to show what Logic's plug-ins are doing to the clips, which may be a little difficult to hear since the parts are individually low in the mix. A few are pretty obvious, like the amp modeling on the bass and the chorus effects on the guitar tracks. My vocals are compressed, and in the case of the bass vocal part on the bridge, pitch-corrected. There is gating or downward expansion on almost every track to minimize background noise from the original recording environment. This is important because, although on an individual track it may be very subtle, the noise floors of all those tracks piled on top of one another adds up, so one wants them as free of background noise as possible. The violins are EQ'ed and have reverb added. The Rhodes piano is compressed to level the dynamics a little bit so it doesn't "pop up" and "pop down" so much relative to the other tracks.
There is more I can't easily illustrate -- for example, the basses, guitars, and Denise's verse vocals have Logic's flex time applied to more tightly align strums, plucks, picks, and consonants with downbeats and with Joe's lead vocal track. This is easy to hear when the tracks are loud against a click or played directly against one another, but that seemed too difficult to get across in this little audio essay, so you'll just have to take my word for it that it tightens up the rhythm.
Finally, I show briefly what the Ozone 4 mastering plug-in is doing as a whole (in a nutshell, it makes the song much louder and allows you to hear all the parts, but also adds a final stage plate reverb, and does harmonic excitation, exaggeration of the stereo image, multi-band compression, and limiting).
The completed song appears at the end. It is down-converted to 16 bit and dithered, and of course turned into a lossy MP3 file, which even at 320K takes a toll on the audio quality, but I hope you can get the drift of it all anyway.
The final song is available on my Bandcamp page here.
The Evolution of a Song MP3 File