Posted
Hello, Andre,
I have experimented with Phrygian Dominant scale in the role of a global (horizontal) key. To this end, I have prepared an alternative palette based solely on Phrygian Dominant. When this scale starts from C it should contain the following tones:
C, D flat, E, F, G, A flat, B flat, C'
Then I have drawn a horizontal figure representing a one-octave upward and downward diatonic run from the base tone. I have played this figure over the global key of C.Phrygian Dominant with the interpretation parameter almost switched-off (but not bypassed, of course), expecting the above mentioned tones to be rendered. However, to my surprise, the upward half of the figure actually rendered the following tones:
C, D flat, E, F, G, B flat, C', C'
So, in the 6th step, A flat was skipped and C' was repeated in the 7th and 8th step instead. On the other hand, the downward run indeed rendered the tones as expected:
C', B flat, A flat, G, F, E, D flat, C
Is this behavior intentional? It doesn't seem right to me since I didn't use the interpretation features and yet, the results are somewhat unpredictable. I have attached all the relevant files for this test: the arrangement, the palette, the modified catalog.
Thanks for your comments in advance!
Miloslav
Sat, 2018-06-23 - 00:17 Permalink
Voice leading avoids weaker notes in favor of strong notes, depending on harmonic context and rhythm. This leads to different runs for the same figure at different places on the time line.
Turn off VL completely and you'll get melodic runs that are agnostic to rhythm and context. This makes sense for some melodies, but not all.
Sat, 2018-06-23 - 18:39 Permalink
Hello, Andre,
thanks for your response but have you tried to open the attached arrangement? I think I have tested every possible setting but the described behaviour cannot be avoided. Bypassing the Voice Leading for all notes didn't help, either. And yes, all the notes are marked as Weak Anchors.
So, I'm afraid this sounds like a bug to me...
Best regards
Miloslav
Tue, 2018-06-26 - 19:28 Permalink
In your arrangement, the figure symbols are not connected (isolated, anchor-only segments). This makes Synfire treat them as if each note is supposed to be starting a new segment (sort of a line break). The results will be different if you connect them.
Depending on which symbol you make the anchor of that connected segment, results will also differ. There's a section in the user manual detailing the importance of anchor and sequence in the "Concept" chapter.
Tue, 2018-06-26 - 22:17 Permalink
Hello, Andre,
you are right: if I group the upward run into one segment and mark its uppermost note as the anchor then the figure finally produces the phrygian dominant scale. So, for me the conclusion is that Voice Leading rules built into Synfire cannot be switched off completely - at least, not for the horizontal symbols.
Thanks for your help!
Miloslav
Wed, 2018-06-27 - 09:59 Permalink
You can disable VL for the segment, sure. There's also a template for Interpretation that has all VL turned off.
On a general side note, there are basically two approaches to rendering music from abstract descriptions:
A) Strictly applying (predictable) math and algorithmic mappings.
B) Using AI to figure out what might be the best way to make the descriptions sound like naturally performed music.
Approach A is very satisfying to the scientifically minded user, at least while you work at it, but often less so for an unsuspecting listener (or when you listen to it the day after). Music happens to be more a product of cultural evolution than beautiful math. If anything, it could probably be modeled by physics, but that would be an entirely new approach.
Approach B takes into account the typical behavior of melodies, harmony and rhythm in human-made music, in order to solve the puzzle of bringing multiple musical expressions together under the rule of any arbitrary harmonic progression. This approach excels at the re-interpretation of existing musical material in a different context.
We thought approach B would be both more practically useful and more unique, as there are already many, many tools available for A.