Re: Automatically realize figured bass

[Richard Shann]

Hi Malte,

Almost all my music typesetting involves typesetting figured bass, and
in addition I have created a way of realizing figured bass by playing on
a MIDI keyboard (see the demo at https://vimeo.com/62426412 ). I have
used this to create LilyPond engraved scores which a keyboard player
found adequate for playing an accompaniment.
One thing I think you would need is the ability to set the likely rate
of chord change - figured bass is an ambiguous notation even when the
figures are relatively complete (and they are rarely that) - but on the
question of whether a note is to be treated as a passing note or a
harmony note it is completely silent.

I guess you could consider using Denemo as your representation of the
data (bass notes with figures attached, and if you want to really push
the boat out, all the other notes that are being sounded at the same
time). It has a scheme interface which allows you to enquire about notes
and their durations, iterate, etc and it does have a separate data type
for the attached figures. (I noticed to my dismay that Finale when
exporting figured bass in MusicXML just exports bits of text distributed
in suggestive manner below the bass note, indistinguishable from any
other text that might be there). I have quite a large collection
(several hundred) of scores with figured bass in Denemo format (and
hence available also in LilyPond format), if they would be of use to
you. They include the entire set of Handels figured bass exercises,
which covers quite a lot of figured bass notation (these are publicly
available in the Denemo distribution itself).

HTH

Richard


On Sat, 2014-04-05 at 16:50 +0200, Malte Meyn wrote:
> Hi everybody,
> 
> a few weeks ago my father asked me whether there is a tool that 
> automatically realizes figured bass (I think you say “realize” for the 
> act of composing the right hand accompaniment, don’t you?). I didn’t 
> know such a tool, so I wrote a simple one in Haskell but this wasn’t 
> clever enough (worked only under certain preconditions, used 
> backtracking instead of something faster, didn’t weight solutions by 
> their “goodness”, etc.). But now I have several ideas how to improve 
> such a tool (represent chords as vertices in a graph, edges are weighted 
> (unison/octave parallels get a very high score etc.), then find the best 
> solution using the Dijsktra or Bellman-Ford wayfinding algorithm, …).
> 
> I don’t know whether such a tool could be implemented in LilyPond itself 
> (as an engraver?) or it would be better to write a separate program 
> (like lilypond-book/musicxml2ly/…). In the latter case, I think we would 
> be relatively free in the choice of a programming language ;)
> 
> The german LilyPond forum doesn’t seem to be very interested in such a 
> project yet. But there are so many more LilyPonders “out there” ;) Would 
> someone on this list be interested
> • to use such a tool if someone implemented it?
> • to implement such a tool? (knowledge of figured bass or a programming 
> language would be good ;)
> • Or is there already such a tool for LilyPond that I don’t know?
> 
> Cheers,
> Malte
> 
> _______________________________________________
> lilypond-user mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/lilypond-user