On standard stuff, ACW seems to work pretty well, coming up with reasonable chords.

However, it seems to have a problem when the comping is pushed or laid back a little, coming up with the most absurd chords. If I manually move the bar start a little so that it lines up with the chord start, then the chord algorithm seems to perform much better. I've tried the tempo adjustments for slightly faster/slower tempos, but the problem is not tempo itself, but relative timing within the bar. I can't seem to find a way to shift all the tempo bars ahead or behind a small amount so the right chords are detected, other than manually moving each bar.

Several possibilities come to mind regarding possible future enhancements to ACW to mitigate this problem. First, it seem that each eighth-note position in the memasure is assigned a note the entire period, yet if I move the bar start just a fraction of the eighth-note, the correspnding notes change dramatically. It seems, then, that the "granularity" of the notes is much finer than it shows, so picking out the notes within a bar shouldn't be constrained to those shown. Second, such rhythmic comping makes new chords at a relatively higher volume than the decaying chords from the previous bar, so choosing the louder notes within each eighth-note might give a better chord choice.

If you could select say 16th note granularity, and if the note bars were colored according to their velocity/amplitute, then by allowing each 16th or eighth column of notes to generate the corresponding chord, the user could click the appropriate column depending on whether the comping was pushing or laying back. Having a way to shift the note columns in every mesure when selecting the 1/2 measure chord, and letting this "comping tempo" be a time-shift in the tempo map, the ACW could perform much better on rhythmic songs.

Anyone figured out a better way to do this without waiting for a new ACW?

Jerry Schneider