Simon Albrecht <address@hidden> writes:
Hello,
I've got a problem: I'm typesetting a Brahms lied where some melismata
are only for one stanza and not for the other. Now look at the
following example (output attached):
\version "2.19.3"
<< \new Staff <<
\new Voice = "one" { \voiceOne c'4 c' c'8 c' c'4 }
\new Voice = "two" { \voiceTwo s4 s c' s }
\new Lyrics \lyricsto "one" {
foo \set associatedVoice = "two"
foo \set associatedVoice = "one"
bar
bar
}
The associated voice has to be set one syllable early, which I did,
only the lyrics "come back" to voice "one" one note early: it should
be for the last quaver and not for the second crotchet. It seems to me
that the roots of this problem lie very deep. An optimal fix would of
course be to make the effect of \set associatedVoice enter immediately
rather than one syllable later, but this is probably far away.
Can anyone think of a workaround?
Just keep the same associated voice throughout for the variant with most
syllables, flipping it between \voiceOne, \voiceTwo and \oneVoice as
needed.
I lean towards not using \lyricsto here but just placing the durations
in the lyrics. That makes the timing unproblematic and robust against
slurs/beams/whatever, saving a lot of fiddling.
Maybe a cute LSR entry would be a Lyric_time_translator which is placed
in a lyricsto-governed context and then writes out the source file with
proper durations in all its lyrics so that you can then use this and
drop the lyricsto.
Then you need to get only one stanza synchronized at a time...
Associated voices still are good for not just aligning on the time but
note values.
In case someone might ask why this is necessary, I attach what I
already have of the entire lied also. The \makeOctaves function is
from LSR 445 <http://lsr.di.unimi.it/LSR/Snippet?id=445> and I added
it to scm/music-functions.scm or a similar one in order to always have
it at hand.
Non-compilable examples are a nuisance.