I observed that there are issues with the chord playback of guitar harmonies when using 9th and 13th chords. These likewise propagate into MIDI exports where the improper chord formations originating in Hookpad are further confirmed within other independent products (PreSonus Studio One, Signal, etc.). In short the 9th chords are playing as incorrect slash chords and the 13th chords are just 11th chords. This is not a matter of voicing as it is understood that a note gets dropped commonly with 13th chords. The issue is that the fundamental notes in the chords do not seem to represent any reasonable voicing for a 9th or 13th chord. The key harmonies (e.g. pianos) seem to all have their reasonable chord voicing intact, so it seems this is just a matter with guitar harmonies at this time. Orchestral harmonies seem to not even apply 9th, 11th, or 13th chords at all (e.g. violin section), so that may be another manifestation of this higher ordered chord issue with some harmonies. Picture attached to highlight an example and where the breakdowns occur.
With guitar chords, we do our best to match the chord voicing the way chords are actually played on a guitar. While this system isn’t perfect, we find that this gives the most realistic picture of how a song/track would sound if played on real instruments. The guitar voicing algorithm works by first checking to see if a given chord can be played in open position using standard fingerings for common open position chords (G, C, D, A, E, em, am, etc). If it doesn’t find it, then it examines the E-barre and A-barre fingerings, and picks the one with the lower barre fret number. If it doesn’t find it among standard barre fingerings then it does its best to come up with a reasonable fingering.
Since C-rooted chords will typically be A-barre fingered, we use: x02100
Which will sound as V over I, and to my knowledge is the most commonly accepted A-barre position for maj9 quality chords on guitar.
An alternative A-barre pattern for maj9 is: x1021x, which would contain the third of the chord, but would only have 4 strings. Might be worth considering though, since it omits the 5th but includes the 3rd, which strictly speaking is a more proper 4-note voicing.
wouldn’t (e.g.)) Cmaj9 look like this:
Hi Ryan,
Thanks for the thoughtful and quick response to my observations! I really should have mentioned previously that I sincerely LOVE using Hookpad and hope that any observations, questions, or feedback I provide are taken constructively. As somebody that comes from a scientific career background while also loving to play guitar, sing, and create music, I find Hookpad is unlike any other tool for drafting compositions given its balance of logical theory foundations and creative flexibility.
I appreciate your overview of the Hookpad algorithm used to determine guitar chord voicing as that at least helps to explain why I observe the particular constructions. In retrospect, calling the Hookpad voicing for guitar harmonies “wrong” might have been unfair on my part as I did recognize the voicing choices contain a subset of the actual chord notes. I think your latter comments really strike at the core issue though.
As a guitarist presented with something like a Cmaj9 chord, my default inclinations are that whatever voicing is ultimately chosen, there is going to be the chord root note (e.g. C), the major note (3rd), and then the additional note (e.g. 9th). Esoteric or subjective alternatives aside, I would argue that the most common (i.e. fundamental in theory) voicing would have those three elements in this example. This is why the Hookpad representation bothered me so much for Cmaj9, as it doesn’t include the 3rd (E) in the formation and so the major/minor foundation of the chord is undefined. Again, creatively we can all drop whatever notes we want for a chord and make things work subjectively, but from a fundamental standpoint for a tool like Hookpad it was a surprising choice (to me at least). So although the current Hookpad voicing maybe isn’t “wrong” for a Cmaj9 guitar chord, that voicing would have been way down the list of options for me (as a guitarist) given multiple options to work in the 3rd and 9th notes (and instead drop the 5th). Ironically one can even play a Cmaj9 with one finger fretted and strumming the top five strings to get all five chord notes…
(Source: www.scales-chords.com)
So the impact of this was tangible as when I was drafting a piece in Hookpad recently I was taken aback that I couldn’t hear any major harmony with my maj9 chord guitar backing choice alongside my melody. I couldn’t quite figure it out at the time and by chance I switched to piano backing and it sounded more like what I was expecting at that time. It wasn’t until days later when I was working in PreSonus Studio One that I found the MIDI chords for the guitar harmony version were not being labeled with the chords I expected and hence motivating my initial bug report. In short, having the arguably uncommon guitar voicing really messed with my ears and creativity as I was expecting something with the foundational notes in there (e.g. root, 3rd, 9th), but got something more unusual (dropping the 3rd).
As I already mentioned, I really enjoy Hookpad. This current limitation really puts a crimp of the utility of the guitar harmonies though, especially if using more of these complex chord selections (9th, 11th, 13th notes). So that’s going to in many cases drive me now to just use piano for the Hookpad compositions and then switch those to guitar later in my production workflow via live guitar tracks or virtual instruments (i.e. in a production DAW like PreSonus). That sorta makes me sad as I enjoy hearing a guitar backing in Hookpad during the initial creation phase, but I find it will be hard now to get over these guitar harmony voicing choices and it certainly would become more and more of a distraction.
I might suggest for consideration one simple solution which is to wholesale change the guitar harmonies to match the piano harmonies note-for-note as the keyboard instruments seem to have very reasonable chord voicing constructions that I would expect. In that regard, the current guitar algorithm to try and look at barre chords and derive notes from that might be overkill or maybe even a misguided endeavor. The current Hookpad piano voicing choices seem to be much closer analogs to what I actually would expect for guitar harmony notes (and maybe arguably one-for-one swaps in most all use cases foreseeable within the bounds of Hookpad).
An explicit chord builder where individual notes are chosen might also be another option for harmony instruments. This sorta exists now, but it doesn’t quite do what I’d expect either (e.g. adding the 13th note checkbox to a maj11 chord on guitar harmony does not actually add the 13th as that underlying algorithm is still overriding the selection and providing another unusual guitar voicing choice).
I’m honored that one of the Hookpad creators took time to respond to my observations. Thank you, Ryan! Hopefully my comments are helpful and considered in making roadmap decisions for the future improvements and evolutionary path of Hookpad!
+1 on the use of the piano note approach for guitar - w/ the idea maybe make it a setting for guitar to use piano chords vs “pure” guitar approach.