Re: Dynamics within divisi

[Heikki Tauriainen]

Hi,

On Fri, 2017-09-15 at 09:42 -0700, SoundsFromSound wrote:
> Alistair Millar wrote
> > When I use the following divisi, I get the error message:
> > "(De)crescendo
> > with
> > unspecified starting volume in MIDI." A \f is set a couple of bars
> > before
> > the
> > divisi bar. 
> > 
> > <<
> > {r8\< c'4-.\> c8-.\!}
> > \\
> > {g4-. c,-.}
> > > > 
> > 
> > How do I overcome this? Thanks
> > 
> > Alistair
> > _______________________________________________
> > lilypond-user mailing list
> > lilypond-user@
> > https://lists.gnu.org/mailman/listinfo/lilypond-user
> 
> Hello,
> 
> Perhaps it's related to this?
> https://code.google.com/archive/p/lilypond/issues/3945

A more up-to date link for this issue [1] shows that this warning was
removed in version 2.19.44, so probably the easiest way to just get rid
of the warning without modifying the input would be to install and use
a newer version in case you don't care about the precise volume of the
divisi notes in MIDI output.  (If you do, read on.)

Even though the warning has been removed, the handling of MIDI volume
at the beginning of new Voice contexts (implicitly created by the \\
construct in the above code) remains the same even in 2.19 versions,
however: without adding explicit initial dynamics to every new Voice,
each new Voice – whose MIDI volume is by default independent of the
MIDI volume of all other voices – will start with the default MIDI
volume.  This means that “[setting] \f [...] a couple of bars before
the divisi bar” has actually no relevance in the above code because
that \f applies to a different Voice, and not any of the Voices created
for the divisi notes.  (Also, \< and \> in the upper divisi notes will
change the MIDI volume only for the upper notes, and not the lower
notes because the upper and lower notes are in two separate Voices.)

(For more details, see comment [2] in the mailing list thread linked to
in the issue – I believe the example about the behavior of MIDI volume
at Voice boundaries is still valid, as are all workarounds suggested in
the comments that follow.)

[1] https://sourceforge.net/p/testlilyissues/issues/3945/
[2] http://lists.gnu.org/archive/html/lilypond-user/2014-05/msg00429.ht
ml

-- 
Heikki Tauriainen