In this example, Dorico decided to read the input from the MIDI keyboard as an F flat, although the same voice then proceeds to F. This leads to three cautionaries being added, which would not be necessary if an E would have been chosen. (The key is set as Eb major, not C minor, but in the latter that F flat would feel even weirder.)
I do know that I can always respell the respective notes, but I would like to understand the software’s rationale why it would prefer to write an F flat in the first place. As I understand it, secondary leading notes much more frequently lead upwards, creating secondary dominants when harmonized, so I am really puzzled by the appearance of a flat 2 (aka “ra” in solfege) in a major scale - which isn’t then corrected even with the progression of the voice and further harmonic context.
In a minor scale, ra would be needed for the Neapolitan sixth chord, which on the other hand is rather uncommon in a major context. In contrast to the (natural or harmonic) minor scale, the major scale offers a triad with a perfect fifth upon the second scale degree, so there should be much more need for a secondary dominant (containing di as its third) leading towards it than for a Neapolitan.
Is there some option to adjust the preferred accidental spelling for a scale? If not, is this something that is thought about for future versions?
P.S. I just tried removing the soprano’s A flat and then reentering the alto to check if the software just wanted to avoid a diminished fourth, but it still read the input as F flat.
As you input the rising line Eb-E-F Dorico starts off by trying to avoid two notes with the same pitch name, so when the second note is first input it will spell it as Eb-Fb. If the very next note was an F, Dorico would indeed go back and respell the Fb as an E natural. However, because there are two Fb/E notes, Dorico assumes that when you come to input the second one, you want the same spelling as the first one, and then it won’t respell two notes back.
The short version is that Dorico doesn’t have a simple fixed spelling table - it’s also looking at note-to-note movement, and that explains the behaviour here. (It also looks at vertical relationships between notes that are input at the same time, but each voice here is monophonic so that’s not relevant.)
1 Like
Ah, I see, thank you for clarifying!
Would it technically be possible to have the spelling algorithm take the context of multiple simultaneous voices (EDIT: that have been input before or may be added afterwards) into the equation?
This would be hard to do during note input, both conceptually and technically, but it is possible to respell notes en masse after input using the Respell Notes Automatically function. This isn’t foolproof either, of course, because fundamentally this is a difficult problem to solve.
Incidentally, you can use the Respell Using Note Name Above/Below functions during MIDI input. So, to return to the previous example, if you had done Eb, then E (spelt as Fb), then respell to E and then hit the E/Fb note again, it would spell it as an E the second time even before you hit the next note (the F), because it keeps track of spellings you’ve explicitly overridden during the same input session.
1 Like
Just tried my hands at this function, which I wasn’t aware of yet. Indeed, it isn’t foolproof: It only changed the second Fb to an E and kept the first one no matter which setting I used. Could it maybe be developed further into some kind of spell checking , which could then take the larger vertical and horizontal context into account and prompting the user afterwards which accidentals they want to have respelled?
EDIT: Your second advice I will follow in future note input sessions, thank you!
