Wow, that's really odd. But I think it's valid, if not rather esoteric. It's basically an altered scale, or a tritone substitution in action. For example, the C7alt scale is built from the scale C, D♭, E♭, F♭, G♭, A♭, B♭. Enharmonically, this is almost the same as the scale for G♭7, which is the tritone substitute of C7: G♭ (=F♯), A♭, B♭, C♭, D♭, E♭ (=D♯), F♭ (=E). The only difference is C, which is the sharp eleventh of the G♭7 chord. Thus, the alt chord is equivalent to the tritone substitution with a sharp eleven alteration. In this setting, the tritone substitution primarily implies a Lydian ♭7, or Lydian dominant scale. In the case of C7 to Bmaj7, the implied scale behind C7 would be C D E F♯ G A B♭. Because of this, the extensions of 9, ♯11 and 13 are all available while the ♯11 is where it shares with the altered scale.
In your case, substitute Eb7 (from F minor/Ab major) and A7 (from B minor/D major) for the C7 and Gb7. So you're basically getting an Eb Lydian Dominant sound, which is the fourth mode of Bb melodic minor. Lydian Dominant can sound very jazzy, so maybe you're picking up on that. However, not hearing that scale over a parent Eb7#11 chord can make it hard to get a feel for it. Without hearing your examples or seeing your settings, I'd be more inclined to think that the harmonizer is varying the notes as needed to stay within the confines of a certain interval relative to the key of B minor - at least that's how my Harmonyman does it. Because if it truly is harmonizing completely in tritones (which is what the interval from B minor to F minor is), everything should sound like crap. Of course, I could be completely wrong. Just my 2 cents.
oke 
