For this thread I'm modeling Autumn Leaves as a 4-section tune as follows:
Section 1: An 8-bar ii-V-I progression in C ma
Section 2: A second nearly identical ii-V-I in C ma
Section 3: An 8-bar ii-V-I in A mi (relative to C)
Section 4: A 2nd, nearly identical ii-Vi-I in A mi
I simply took an example of a 'standard' 32-bar jazz progression out of a book, then bounced it against my Autumn Leaves melody in C. It seems mostly okay, except for the following problems.
1. Sect 1 bar 8 is an A7, but the melody contains a C natural
2. Sect 2 bar 6 is an E7, but the melody contains a C natural
3. Sect 2 bar 8 is an Am7, but melody has a G#
4. Sect 3 bar 2 is an E7, but melody has A natural
5. Sect 3 bar 4 is an A7, but melody has C natural (see 1. above)
6. Sect 4 bar 2 is E7, but melody has F natural
7. Sect 4 bar 3 has a G#7, melody has E natural
8. Sect 4 bar 4 has an F#7, melody has an A natural
Item 3 seems fixable by sub'ing an Amb7
Item 4 seems fixable by sub'ing an E7sus.
But I guess what I'd like to know is how to approach this methodically. It doesn't seem reasonable that I really want to eliminate all discordance, and I may be assuming that any two notes a half-step apart are going to sound horrible. I just cant find any way to stay within the progression and still resolve all the items above.
Maybe my first and biggest problem is not understanding the logic behind tension and resolution from chord to chord when you have one chord per bar--in other words, in a tune structured this way, what is the likely overall pattern of resting, mildly active and very active throughout the piece?
I now feel I'm rambling--do these questions make sense?