Re: Architecture for Fret Diagrams

[Carl Sorensen]

On Fri, 2004-07-23 at 10:06, address@hidden wrote:
> As I said before, I really do love the quality of Lilypond output. So I hope
> you'll take this in the spirit it's intended: observations from a domain
> expert
> that may or may not be useful to you in the future.
> 
Thanks for the input.  I'm just a relatively poor hobby guitarist.  I'm
interested in lilypond as a way to get simple fret diagrams for simple
tunes.

I've created a fret diagram markup for version 2.3.x (I think it started
as part of the distribution for 2.3.4, but I may be wrong.).  As it
stands now, you can put fret diagrams into your music in a very
straightforward way -- almost easy.  However, it's only a markup, rather
than part of the music.

> Agreed. It's a difficult problem. As a point of comparison, I ended up
> buying a copy
> Finale, which does chords reasonably well. There are over 200 chord forms
> in three major variants, rendering on three different typefaces in Finale
> (one
> American system, and two jazz systems, each of which has multiple
> Major/minor systems). Even with a library of 200+ chord forms, I still
> find myself  creating new chord forms from scratch. Really common chords in
> modern jazz repertoire such as Maj13, and sus b9, sus13 are not present. And
> my preferred Major/Minor system (delta/m) isn't fully supported in either of
> the
> jazz font systems.
> 
> The nice thing about having all those chord forms at your fingertips is that
> you can choose chord forms that are appropriate for the chart. e.g.
> 50s jazz charts tend to use   F7(b5); more recent charts use F7(#11) to
> mean the same thing (the latter being much more accurate). When I'm
> writing charts, I tend to respect the historical context when choosing chord
> forms.
> 

Right now, I really don't want to jump into this complexity.  I'm just
trying to get some simple fret diagrams included with the music, with
the idea of being able to layout simple (i.e. 3-5 chord) songs to help
beginning guitarists.  So I guess that's where my emphasis has been.


> 
> The benefits of a markup-based chord notation system: users can set chords
> anyway they want, even
> mixing conventions in the same chart. It also anticipates scalar chordnames
> which are common in
> the jazz repertoire, such as F.'Lydian' (used a lot in 1960s modal charts).
> 

I think you'll find now that the lilypond markup system is up to the
task of doing a markup-based chord notation system.  But it won't
generate the MIDI output you might want to be able to proof things.


> > \frets
> 
> Probably not quite right. The problem:  chord names and fret diagrams should
> be a single element for
> layout purposes. If you use \frets, then chords and fret diagrams would be
> set in separate contexts (?),
> in separate parts of the input, and would introduce the possibility of fret
> diagrams without chords.

My initial thought was also that chord names and fret diagrams should be
part of a single element.  Han-Wen has thought about the logical
structure of lilypond a lot more than I have, so I tend to defer to him.

> 
> Users would have to do a fair bit of work to keep fret diagrams and chord
> names aligned
> properly..

As long as both the fretdiagram and chordname engravers handle the same
music expression, I don't see any difficulty at all keeping things
aligned.  Of course, if you're going to pass the same music expression
to both engravers, it might be preferred to have a single context that
includes both.

> 
> Better would be to extend the \chords syntax to include an independent fret
> diagram notation as part
> of the chord name, I think.  Something like:
> 
>         Bf.13[vi 6-1,X,6-1,7-2,8-3,X]
> 
> (or whatever the proposed fret notation is).

I think that this is easy syntactically, but it's inconsistent with the
way lilypond handles music events.  It could be done, but I doubt that
it's in my immediate plans.

> 
> The basic point that I wanted to make that recycling notes from
> the chord name part of a \chord element is probably not a good idea.
> 

I believe I agree with you here.  But it's not a bad idea at all to have
notes with string and finger notations that can be automatically turned
into fret diagrams.  That is, instead of setting the chord by name, with
a fret-diagram string added to it, I think the intended architecture is
likely to prefer setting the chord by note, string, and fingering, then
have the name automatically determined by the chord namer.  I guess the
quality will depend on how well the chord namer works, and I'm not the
person to try to explore that.  I don't know enough about the different
chord naming systems.

Thanks for your input.  I'd appreciate your getting the current version
of lilypond and trying out the fretdiagram markups.  I'd like some
feedback on how the system might be improved.

Thanks,

Carl Sorensen

>