Transposing staves with chord diagrams

• Dec 11, 2020 - 21:54

I usually have no problem transposing from one key to another. The notes and the chord names always change to the new key. But now I am trying out chord diagrams above the staves and they are not transposing. The notes on the staves do transpose, but the chord diagrams and symbols do not. I wonder if there is a problem with how I've notated the diagrams? But I've tested it by selecting the first measure (which I know is correctly symboled as simply G and C, but that measure too does not transpose in the digrams or the symbols. What else could be the problem?


Chord diagrams don't transpose. You will need to changes those manually. There are many ways to play most chords on an instrument and an automated method would be hit and miss in giving acceptable results.

This issue is about 90% resolvable. This is evident when we break the problem down into it´s components.

1) Chord symbols do transpose without the diagrams and it was the introduction of the diagram that brought this issue with it. Therefore, that issue need not exist.

2) Break down the chord diagrams into two classes, stock diagrams and customized diagrams. Stock diagrams very rarely reflect the voicing in the staff and are transposable. Any songbook will show us that the vast majority of the diagrams are stock. The only manual work would be on the custom diagrams.

Generally, the voicing is indicated by the symbol, such as C/G. The stock diagram is only a hint at the chord structure.

In reply to by Rockhoven

Break down the chord diagrams into two classes, stock diagrams and customized diagrams.
Any songbook will show us that the vast majority of the diagrams are stock.

Are "stock" diagrams to be restricted to major, minor, and seventh chords? Maybe include aug, dim, maj7, mi7 chords? Where exactly do songbooks draw the line?

Also, who will this benefit?
At the songbook level of using "stock" diagrams, guitarists are more likely to "capo up" to change keys, so they can follow the same diagrams (rather than having to change to different chord fingerings).
I've witnessed this at many coffeehouse and open mic venues... :-)

To maintain a library of "stock" or "basic songbook" diagrams linked to chord symbols which could be invoked upon transposition seems unnecessary, especially since the vast majority of chord diagrams exist outside of popular songbooks and so would not be included. You call them "customized" but 6th chords, 13th, minor 6th, 6/9, etc. also have "chord shapes".
If "stock" chord fretboard diagrams are transposed and "customized" diagrams are not transposed, a score will lack uniformity.

The best idea I've seen has to do with voicing a chord using its fretboard diagram (sort of like reading the TAB fret/string) and then using that information to transpose - and transfer the new fret/string info. onto a 'transposed' fretboard diagram. This would work for any fretboard diagram, with no internal library of "basic chords" needed.

P.S.: The topic occasionally surfaces...
Even if chord symbols coupled to fretboard diagrams were able to be transposed (e.g., for 'stock' chords), it might result in mismatched chord symbols and fretboard diagrams (for 'non-stock' chords).

In reply to by Rockhoven

Chord symbols attached to fret diagrams don't transpose indeed. I'd be inclined to call that a bug, except that since the fret diagrams don't transpose either, if the chord symbols did transpose, that would arguably be worse.

As for automatically transposing chord symbols, to me that's not really feasible. I mean, sure, we could take the position that some appropriately small handful of common chords could get transposed, but then what happens if your song uses some of those combined with some we don't handle? Again, I think the result would be worse than doing nothing.

When you say "any songbook will show us that the vast majority of the diagrams are stock", I guess you mean, basic beginner songbooks for basic beginner songs. But if you look at the wider world, I think you'll find quite the opposite.

OK. Taking the problem in order.

1) The notes in the staves have always transposed along with the chord symbols. What justification is there for disconnecting the chord symbol transposition from the stave transposition? The transpostion of chord symbols serves as a guide for manually inserting the transposed chord diagram. Let´s get that back into working order, so that it transposes the chord symbols along with the notes in the staves. Unless you want to disconnect chord symbol transposition from the transposition of the stave, I see no reason for this issue.

In reply to by Rockhoven

What justification is there for disconnecting the chord symbol transposition from the stave transposition?

Chord symbol transposition works for scores without fretboard diagrams. So not disconnected in this case.

For scores containing fretboard diagrams...
It would mean that the chord symbol itself wouldn't match the fret diagram shown, unless the fret diagram gets changed.

The transpostion of chord symbols serves as a guide for manually inserting the transposed chord diagram.

Yes, so prior to entering fretboard diagrams, transpose the score containing chord symbols, and then use the transposed chord symbols "as a guide for manually inserting the transposed chord diagrams".

This issue is nearly fully resovable. Break the problem down into smaller problems. You can solve the parts of this problem in about an hour of intermittent thinking time, over the course of a few weeks. There are about 3 or 4 stages to resoluution. See if you can get through the first stage. I'll check back next month.

Well, in the past month, I managed to pick up my guitar and see that any chord can be transposed with it's voicing intact. So I don't understand why the guitar cannot be transposed. The fact is that I can select and transpose any guitar chord, just as I can select and transpose a piano chord. Then the chord symbol should also transpose, just as it does for piano.

I found three possible avenues for a solution, if voicing is a legitimate concern. But I don't want you to miss out on the joy of discovery. Pick nup your guitar and start in on the joy of transposition. How did you do it? Why does not our musescore program transpose guitar chords and the chord symbols?

In reply to by Rockhoven

I guess you mean transposed "up" with voicing intact, by using barre chords. Yes, this is often possible (although eventually you may run out of frets or fingers). But the real issues are that 1) transposing down this way isn't possible at all, and 2) most people would not normally choose to place a C chord as simply an A chord barred up three frets. So it's just not something of practical value most of the time. Instead, you need to use your own intelligence to choose an appropriate voicing, Not saying someone couldn't program in some AI into MuseScore for this, but it's not on the immediate roadmap.

In reply to by Marc Sabatella

You wrote:
2) most people would normally choose to place a C chord as simply an A chord barred up three frets.
You mean like this?

Are you sure about that? Especially when it concerns people who need fretboard diagrams?
I would think they would prefer the good old "stock" (Rockhoven's term) songbook diagram:

In reply to by Marc Sabatella

OK. Here's the thing. I can transpose a flute out of it's range. Same with every other instrument, in my experience. What's the deal with guitar? If I want to transpose it down theoretically, it should be available. I may want to tune the guitar down a whole step or more, but it's nobody's business if I do. This is Third World thinking (Oh, YOU can't do THAT!) Anyway, it should have the same theoretical range as the practical range allowed any other instrument.

And if I apply the barre in the diagram, the chord symbol should change. If not, leave the symbols unattached to the diagram because if I have a book I've made of 1000 chords for guitar players, I shouldn't have to go through and manually reenter all of the symbols on ntop of having to redo all the chord fingerings. Have a Clear the diagrams and the symbols. Just leave blank diagrams for filling in.

But there are several interesting ways to think about this and develop this feature.

In reply to by Rockhoven

No one ever said it's impossible. Just that it's impossible to always produce the expected results, because the design of the guitar involves changing chord shapes, not merely transposing them. Merely transposing is simply - add barres. But you wouldn't like that, nor would anyone else.

Making the chord symbol change automatically when you alter a fret diagram - or vice versa - is similar possible if you don't care if anyone actually likes the results, but the reality is, this is AI., and no two expert musicians will agree on what to call some arbitrary collection of notes nor will they agree on the best way to fret a given chord smybol. if someone wants to embark on an AI project to make guesses, I fully support them. But it's important to realize that this is absolutely qualitatively different from simply transposition. Transposition is an exact science. Translation between chord symbols and fret diagrams is an art requiring human-level expertise.

In reply to by Rockhoven

I don't understand the question. uitar is transposed in precisely the same way piano is - the notes and chord symbols, which are the things that can be transposed in a very objectively correct way. We have transposition because it is an easily quantifiable action one can write a completely determinsitic algorithm to do, and it's musically useful. We don't have a lot of features to do AI, however - no feature to automatically complete a piece for you in the style of Mozart, no feature to automatically orchestrate a piano piece for brass octect, no feature to suggest lyrics to you for your songs, etc. Not saying these couldn't be fun research projects, but we generally stick to things that are quantifiable, deterministic, and musically useful.

In reply to by Marc Sabatella

That's not the issue. Yes, we have transposition of the guitar and the chords as they are ventered with the K key, but the diagrams don't transpose and neither do the chord symbols attached to them. And this can be improved. We just need to flood this forum with enough guitarists who want it.

We could have complete transposition, including voicing, up the neck, down the neck (below the nut) and across the neck (transformation of chord forms.)

In reply to by cadiz1

I don't know what that means "closing by design." But this is from that same thread:

"MS 3.2.3. Transposition (throught the Transpose dialog) does not work for chord symbols which are coupled to fretboard diagrams—only for uncoupled chord symbols. Although, fretboard diagrams have to be edited manually after transposition, it adds a great deal of extra work if transposition does not work for the associated chord symbols."

And Marc prioritized this a year ago as "P1-High." Then it's explained as not being a bug but a feature. So that is what I'm speaking about. The chord symbols of the fretboard diagrams can be (or ought to be, meaning it is within the realm of possiblity) transposed, because I've written the symbol, and the program need only transpose what symbol I've written, and does not respell the chord. To transpose the score or one chord from C to G we would get that result by transposing up a 5th or down a 4th. The result would be G, not Em6 (C) to Bm6 (G.) We are not asking for the program to interpret the chords. The chords are spelled out clearly by the transcriber.

First, we need simple transposition up the neck, which the current program can be adapted to for the diagrams. Voicing remains the same because the chords are only barred. You move the barre, the symbol transposes and you can view this happening as you edit.

Then there are some other more intricate approaches to this that could allow for transposition across the courses of strings - from one chord shape or form to another. Like the transposition from the open C form to a barred C on the 3rd fret. This kind of transposition involves voicing, but there are solutions. I am saying that we should move this feature forward a little bit at first and tackle the more complicated issues later. But I believe we can achieve full and complete transposition. I just don't want to get way out there on that issue of voicing yet and cloud up what can more clearly be initially understood. Besides, if there is no interest in this, I don't have time to explain the details of how to do the higher class of transposition.

In reply to by Rockhoven

You can indeed flood the forum with requests for an AI feature to try to come up with an appropriate fret diagram for each chord symbol. And then it might indeed happen. It's just absolutely crucial one understands the difference here. Ordinary transposition is a completely deterministic procedure that has one single correct answer. It's like asking a computer to calculate 1+1 - there is never any doubt that the answer is 2. Whereas determining appropriate fret diagrams for any given chord is a complex AI problem with no single solution, just different opinions on which diagram might be preferred by which musicians in which contexts. It's not at all like 1+1 where there is a single correct answer. It's more like asking the computer to devise an equation whose answer is 2 - tons of possible answers, with no way to pick one as better than any other.

Perhaps we could kill 100 birds with one boulder. Some questions first? Can I place multiple dots on the diagram and represent a scale of 8 notes? Or do diagrams only work with chords?

Marc - I think that you are saying that there has to be value for the labour. I agree.

Here's the thing. Since the chords and symbols are already set up and ready for transposition in the score, and there is code that transposes chord symbols normally on the stave, that code should be used to transpose the chord symbols on the diagrams. Because that is one thing done - rather than having to transpose both the chord diagrams and the symbols. The transposed symbols are then a guide for the manual transposition of the chords. This saves a lot of wasted labour on the part of the transcriber.

But you are referring to the labour input of the developers. What about transposable diagrams for all instruments? Let's work out a system for any one voiced instrument. Any one voiced instrument can have a finger diagram. I feel confident that we are going to make some interesting discoveries. So, this would be linear transposition. The steps W S W W S W W would transpose in that order, as would a chromatic scale. Enharmonic equivalents would be handled the same way as on the stave. It sharps while ascending and flats while descending, right? And you can correct the details manually.

In reply to by Jojo-Schmitz

Well, that's one way to look at it. But if I am going to transpose 100 diagrams and symbols, I will have to take note of every symbol and transpose and transcribe each separately into the symbol fields to use as a guide for the transposition of the diagrams. Therefore, the code that transposes the symbols on the stave should be applied to the the symbols on the diagrams, so the transcriber does not have to transpose and transcribe the symbols on the diagrams. Then the transposed symbols become the guide for transposing the chord diagrams.

In reply to by Rockhoven

I"m not just saying there has to be value - no one disagrees that AI has value. I'm saying someone needs to actually propose a design or the AI algorithm to control this, and then there needs to be agreement it produces satisfactory results. Because again, it's not a simple transposition.

If you care to describe in full detail how such an AI algorithm should work, someone can begin to try to implement it. Be sure your proposed algorithm precisely shows how each and every possible fret diagram would be recalcuated after transposition, not just the dozen or so common open chord voicings. Until then, there really isn't any point in spending time on this thread. It is the one single thing that could be done to advance the discussion. Design and propose your AI algorithm, then we can continue talking.

In reply to by Marc Sabatella

I suggest a little step:
There are now 3 default types for Chord Diagrams: Major, Minor and Dominant 7
based on the C-Major scale (Notes C D E F G A B)
If we create some defaults also for the missing 1/2 steps (Notes C# D# F# G# A#) then Chord Diagrams could be transposed. All customizations or unknown Chord Symbols could get empty diagrams.

This is not the ultimate solution but better than no transposition at all.

(This is the solution I went for in the LibreOffice Extensions ChordDiagrams and ChordTransposer)

Do you still have an unanswered question? Please log in first to post your question.