On 25 Jan 2010 at 11:44, Eric Fiedler wrote: > I hear what you're saying...and sometimes a local solution is the > best and only way to go in a pinch. But on the other hand, there's > something about a score with a number of different spacing algorithms > that doesn't look quite right to me. As I understand it, the old > engravers of yesteryear used one set of spacings for a whole piece or > movement, which _did_ require a lot of calculations before hammering > in the first note punch, but which produced results which are easy to > read and in addition exude a wonderful kind of harmonious "rightness" > that has a lot to do with the overall graphical balance between black > and white on the page. Or am I being too fussy here? I'd be > interested in your take on the subject - or anybody's take, for that > matter.
My guess is that however strict any house style that's implemented by human engravers might be, there's going to be a ton of minor ad hoc adjustments that are either not specifically delineated by the house style's spacing rules, or that violate them in any particular case. Computers are not smart enough to do anything but follow the rules slavishly. If you want them to break the rules, you have to define a rule! That is, the computer can never do anything other than exactly what it's told (though introducing a randomness factor, as with Human Playback, might potentially improve results for certain kinds of operations without needing to define rules that override other rules). To really implement what a human being does, you'd probably need to write a computer program that would take too long to calculate the layout. One computer approach might be to use a multi-pass approach, i.e., do a default spacing, then look at it a second time to see if there are any suspicious things that need to be fixed. I'm not sure how easy it is to examine the Finale data for that, but it might be possible to fix certain kinds of errors in a second pass (or flag them and ask the user what you want to do about them). Yet another approach could be to train the spacing mechanism to fix certain things (similar to the way you train an OCR or voice recognition program). However, there can always be unexpected results from that, so whether you wanted the spacing definition to learn something like that permanently, or learn it only for the present file would be a difficult question. Another thing to consider is having your music spacing settings to always incorporate manual changes, but that can lead to other problems (you'd have to turn it off if you wanted to re-run the spacing to get rid of mistaken manual spacing). But I think a human being will *always* do better than a computer. However, the issue is how much time it takes the human being, and if the computer can get 99.8% of it right and the human can fix the remaining .2% in a short period of time (especially when determining how to fix that .2% is not susceptible to algorithmic description, or has no solution that experienced engravers can all agree on), there's little reason to overcomplicate the computer algorithm. There are a few things that Finale always gets wrong and those are fixable in the algorithm, but the things that go wrong only in relatively exotic circumstances are probably never going to be fixed. -- David W. Fenton http://dfenton.com David Fenton Associates http://dfenton.com/DFA/ _______________________________________________ Finale mailing list Finale@shsu.edu http://lists.shsu.edu/mailman/listinfo/finale
