Re: Crescendos, sforzandos, and "impossible" MIDI volumes

[Simon Albrecht]

Am 27.06.2014 20:42, schrieb Knute Snortum:

I want to start a discussion about this and maybe get a change into LilyPond.

Currently, you can get an "impossible (de)crescendo" error in LilyPond if your crescendo has a \sf in the middle [1].  It stops the crescendo and if you restart it, ending in any dynamic mark, you get an error.  This is because a \sf has a MIDI volume of 100% and this volume continues afterward.

My understanding of a sforzando is that it is an accent.  So my feeling is that it shouldn't act like a dynamic marking.  The MIDI volume of 100% is fine for the note this marking is on, but then the volume should go back to where it was before the sforzando and the crescendo should continue.

I agree that the MIDI volume should go back to the previous value (whatever that is…) after \sf. This correctly reflects its musical meaning for what I think. However, I don’t think it is wrong for \sf to stop a (de)crescendo. In any case, it makes much more sense to me to start another one after the \sf than for a (de)crescendo to continue “through” the \sf. So it’s essentially good for \sf to be interpreted as dynamic mark.
I am getting the impression that we are reaching the boundaries of an automatic, numerical determination of the intended dynamic values. In other words, ambiguities arise, which are no problem for a human interpreter, who may then decide in accordance with the context (for better or worse…), and needn’t be eradicated, as I already stated in a similar discussion.
What do you think of defining \sf so that it plays only the affected note, say, 30% louder than the surroundings? Which leads to problems if the previous volume isn’t precisely defined, as in Knute’s example:

  \relative c' {

    c4 \p \cresc d e f |

    g a b c |
% crescendo arriving at 70% ?

    d \sf \cresc c b a |
% d played at 100%, the next crescendo starting again from 70% ?

    g f e d \f |

    c1 

  }

(Note that I have no idea how this is currently or might in future be implemented…)

Just my 2cts,
Simon